AI Governance Glossary (Embed) \u2013 Deepwraite<\/title><br \/>\n<\/p>\n<style>body{margin:0;background:#0b0e11}<\/style>\n<p><\/head><br \/>\n<body data-rsssl=1><\/p>\n<p><\/p>\n<style>\n.dwg-glossar{--dwg-teal:#0ee2cf;--dwg-teal-deep:#0a8f83;--dwg-card:rgba(255,255,255,.035);\n --dwg-line:rgba(255,255,255,.10);--dwg-ink:#e9eef3;--dwg-muted:#8b97a3;\n font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;\n color:var(--dwg-ink);box-sizing:border-box;width:100%;max-width:1340px;margin:28px auto;position:relative;isolation:isolate;\n border:1px solid rgba(255,255,255,.12);border-radius:16px;\n background:radial-gradient(120% 120% at 88% 8%,rgba(14,226,207,.10),transparent 48%),#111110;\n padding:clamp(26px,4vw,40px) 28px;}\n.dwg-glossar::before{content:\"\";position:absolute;inset:0;z-index:-1;border-radius:16px;\n background-image:radial-gradient(#FFFFFF0D 1px,transparent 1px);background-size:22px 22px;\n -webkit-mask-image:radial-gradient(120% 80% at 50% 0%,#000 25%,transparent 72%);\n mask-image:radial-gradient(120% 80% at 50% 0%,#000 25%,transparent 72%);}\n.dwg-glossar *{box-sizing:border-box}\n.dwg-glossar .dwg-brand{margin:0 0 10px;font-weight:800;letter-spacing:.06em;color:var(--dwg-teal);font-size:12.5px;text-transform:uppercase}\n.dwg-glossar .dwg-h1{font-size:30px;line-height:1.18;margin:0 0 12px;font-weight:800;letter-spacing:-.01em;color:#fff}\n.dwg-glossar .dwg-lead{color:var(--dwg-muted);max-width:70ch;margin:0 0 4px;font-size:16px;line-height:1.6}\n.dwg-glossar .dwg-controls{position:sticky;top:0;padding:16px 0 12px;margin:24px 0 6px;\n background:linear-gradient(#111110 72%,rgba(17,17,16,0));z-index:4;border-bottom:1px solid var(--dwg-line)}\n.dwg-glossar .dwg-q{width:100%;padding:13px 15px;border:1px solid var(--dwg-line);border-radius:12px;font:inherit;\n font-size:16px;background:rgba(255,255,255,.04);color:var(--dwg-ink);outline:none}\n.dwg-glossar .dwg-q::placeholder{color:#6a7681}\n.dwg-glossar .dwg-q:focus{border-color:var(--dwg-teal-deep);box-shadow:0 0 0 3px rgba(14,226,207,.18)}\n.dwg-glossar .dwg-chips{display:flex;flex-wrap:wrap;gap:8px;margin-top:12px}\n.dwg-glossar .dwg-chip{border:1px solid var(--dwg-line);background:rgba(255,255,255,.03);color:var(--dwg-muted);\n padding:7px 14px;border-radius:999px;font-size:13px;font-weight:600;cursor:pointer;user-select:none;transition:.15s}\n.dwg-glossar .dwg-chip:hover{border-color:var(--dwg-teal-deep);color:var(--dwg-ink)}\n.dwg-glossar .dwg-chip[aria-pressed=\"true\"]{background:var(--dwg-teal);border-color:var(--dwg-teal);color:#08211f}\n.dwg-glossar .dwg-count{color:var(--dwg-muted);font-size:13px;margin:2px 2px 20px}\n.dwg-glossar .dwg-cluster{margin:0 0 34px}\n.dwg-glossar .dwg-cluster>h3{font-size:19px;margin:0 0 14px;padding-left:12px;border-left:3px solid var(--dwg-teal);\n letter-spacing:-.01em;color:#fff;font-weight:700}\n.dwg-glossar dl{margin:0;display:grid;grid-template-columns:1fr;gap:12px}\n.dwg-glossar .dwg-term{background:var(--dwg-card);border:1px solid var(--dwg-line);border-radius:14px;\n padding:16px 18px;scroll-margin-top:120px;transition:border-color .15s}\n.dwg-glossar .dwg-term:hover{border-color:rgba(14,226,207,.35)}\n.dwg-glossar .dwg-term dt{font-weight:700;font-size:16.5px;margin:0 0 5px}\n.dwg-glossar .dwg-term dt a{color:#fff;text-decoration:none}\n.dwg-glossar .dwg-term dt a:hover{color:var(--dwg-teal)}\n.dwg-glossar .dwg-term dd{margin:0;color:#c3ccd4;font-size:15px}\n.dwg-glossar mark{background:rgba(14,226,207,.30);color:#fff;border-radius:3px;padding:0 1px}\n.dwg-glossar .dwg-empty{display:none;color:var(--dwg-muted);padding:30px 4px}\n.dwg-glossar .dwg-foot{border-top:1px solid var(--dwg-line);margin-top:38px;padding-top:16px;\n color:var(--dwg-muted);font-size:12px;text-align:center}\n@media(min-width:720px){.dwg-glossar dl{grid-template-columns:1fr 1fr}.dwg-glossar .dwg-h1{font-size:37px}}\n<\/style>\n<section class=\"dwg-glossar\" role=\"region\" aria-label=\"AI Governance Glossary\">\n<p class=\"dwg-brand\">AI governance knowledge<\/p>\n<h2 class=\"dwg-h1\">Governance Glossary<\/h2>\n<p class=\"dwg-lead\">The key terms around introducing an AI management system based on <strong>ISO\/IEC 42001<\/strong>.<\/p>\n<div class=\"dwg-controls\">\n <input class=\"dwg-q\" type=\"search\" placeholder=\"Search term \u2026 (e.g. AIMS, high-risk, token, sovereign)\" aria-label=\"Search glossary\"><\/p>\n<div class=\"dwg-chips\">\n <span class=\"dwg-chip\" data-f=\"all\" aria-pressed=\"true\">All<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"aiact\" aria-pressed=\"false\">EU AI Act<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"iso\" aria-pressed=\"false\">ISO 42001 & governance<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"tech\" aria-pressed=\"false\">AI technology<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"cloud\" aria-pressed=\"false\">Cloud & sovereignty<\/span>\n <\/div>\n<\/p><\/div>\n<div class=\"dwg-count\"><\/div>\n<div class=\"dwg-cluster\" data-c=\"aiact\">\n<h3>EU AI Act<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-eu-ai-act\">\n<dt><a href=\"#dwg-eu-ai-act\">EU AI Act (AI Regulation)<\/a><\/dt>\n<dd>The EU regulation (Reg. 2024\/1689) that regulates AI systems by their risk \u2013 the first comprehensive AI regulation worldwide. It applies in stages and imposes obligations according to risk class.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-risikoklassen\">\n<dt><a href=\"#dwg-risikoklassen\">Risk classes<\/a><\/dt>\n<dd>The AI Act divides AI into four tiers: <em>prohibited<\/em> (Art. 5), <em>high-risk<\/em> (via Annex I or III, Art. 6), <em>limited risk<\/em> (transparency obligation, Art. 50) and <em>minimal risk<\/em>. The obligations follow from the classification.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-hochrisiko\">\n<dt><a href=\"#dwg-hochrisiko\">High-risk AI (Art. 6, Annex I & III)<\/a><\/dt>\n<dd>An AI system is high-risk via <strong>two routes<\/strong> (Art. 6): the <strong>Annex I route<\/strong> \u2013 it is a safety component of a product, or itself a product under the EU harmonisation legislation in Annex I (e.g. machinery, medical devices), and requires third-party conformity assessment; or the <strong>Annex III route<\/strong> \u2013 it falls into an area listed there (e.g. employment, education, critical infrastructure, creditworthiness, law enforcement). Consequence: strict obligations such as risk management, technical documentation and human oversight. Exception (Art. 6(3)): Annex III systems without significant risk may be exempt \u2013 but <em>profiling<\/em> always remains high-risk.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-anhang-iii\">\n<dt><a href=\"#dwg-anhang-iii\">Annex III (high-risk areas)<\/a><\/dt>\n<dd>The list of stand-alone high-risk applications in the AI Act: biometrics, critical infrastructure, education, employment, access to essential (including private) services \u2013 incl. creditworthiness and insurance \u2013, law enforcement, migration\/border control, and justice and democratic processes.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-anbieter\">\n<dt><a href=\"#dwg-anbieter\">Provider<\/a><\/dt>\n<dd>Whoever develops an AI system and places it on the market under their own name. Carries the main obligations of the AI Act (conformity, documentation, reporting).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-betreiber\">\n<dt><a href=\"#dwg-betreiber\">Deployer<\/a><\/dt>\n<dd>Whoever uses an AI system under their own responsibility. Must, among other things, ensure human oversight and comply with the provider’s instructions.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-gpai\">\n<dt><a href=\"#dwg-gpai\">GPAI (General-Purpose AI)<\/a><\/dt>\n<dd>General-purpose AI models, such as large language models. Providers of such foundation models are subject to their own transparency and documentation obligations.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-art-50\">\n<dt><a href=\"#dwg-art-50\">Art. 50 \u2013 Transparency obligation<\/a><\/dt>\n<dd>People must be able to recognise that they are interacting with AI or seeing AI-generated content (labelling obligation) \u2013 e.g. an AI announcement at the start of a call or a note on a chatbot.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-verbotene-praktiken\">\n<dt><a href=\"#dwg-verbotene-praktiken\">Prohibited practices (Art. 5)<\/a><\/dt>\n<dd>AI applications the AI Act prohibits \u2013 such as social scoring by the state or manipulative systems that deliberately exploit weaknesses.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-konformitaetsbewertung\">\n<dt><a href=\"#dwg-konformitaetsbewertung\">Conformity assessment<\/a><\/dt>\n<dd>The verification procedure by which \u2013 especially for high-risk AI \u2013 it is demonstrated before market launch that the requirements of the AI Act are met.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-ki-kompetenz\">\n<dt><a href=\"#dwg-ki-kompetenz\">AI literacy (Art. 4)<\/a><\/dt>\n<dd>Providers and deployers are expected to foster sufficient AI literacy among their staff \u2013 a basic understanding of the opportunities, risks and correct handling of AI. The duty has applied since February 2025. In July 2026 the Digital Omnibus changed the standard from \u201censure\u201d to \u201csupport and promote\u201d: the requirement remains, the degree of obligation is softer. Independently of this, an AIMS under ISO\/IEC 42001 requires competence and awareness in clauses 7.2 and 7.3 \u2013 there the evidence is still expected.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-fria\">\n<dt><a href=\"#dwg-fria\">Fundamental-rights impact assessment (FRIA, Art. 27)<\/a><\/dt>\n<dd>Certain deployers of high-risk AI must assess the impact on the fundamental rights of affected persons before use (Fundamental Rights Impact Assessment).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-post-market\">\n<dt><a href=\"#dwg-post-market\">Post-market monitoring (Art. 72)<\/a><\/dt>\n<dd>Providers of high-risk AI must actively monitor their system after placing it on the market and systematically evaluate findings.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-vorfallmeldung\">\n<dt><a href=\"#dwg-vorfallmeldung\">Serious incidents (Art. 73)<\/a><\/dt>\n<dd>Providers must report serious incidents and malfunctions of their AI to the authorities \u2013 within defined deadlines.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-ce\">\n<dt><a href=\"#dwg-ce\">CE marking & declaration of conformity<\/a><\/dt>\n<dd>With the EU declaration of conformity and CE marking, the provider confirms that a high-risk AI system meets the requirements and may be placed on the market.<\/dd>\n<\/div>\n<\/dl><\/div>\n<div class=\"dwg-cluster\" data-c=\"iso\">\n<h3>ISO\/IEC 42001 & governance<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-iso-42001\">\n<dt><a href=\"#dwg-iso-42001\">ISO\/IEC 42001 (AIMS)<\/a><\/dt>\n<dd>The international standard for an AI management system: it defines how an organisation governs, documents and continuously improves AI responsibly. Certifiable \u2013 Deepwraite\u00ae’s core field.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-aims\">\n<dt><a href=\"#dwg-aims\">AIMS (AI Management System)<\/a><\/dt>\n<dd>The living management system for AI: roles, policies, risks, controls and evidence across an organisation’s entire use of AI \u2013 not a document, but a process.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-kik\">\n<dt><a href=\"#dwg-kik\">AI coordinator (KIK)<\/a><\/dt>\n<dd>The central operational role in the AIMS: coordinates set-up and operation, maintains the documentation, manages risks and approvals (ISO\/IEC 42001, Clause 5.3).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-aisia\">\n<dt><a href=\"#dwg-aisia\">AISIA (AI System Impact Assessment)<\/a><\/dt>\n<dd>The assessment of an AI system’s impact on people, groups and fundamental rights (ISO\/IEC 42001, Clause 6.1.4) \u2013 the AI-specific complement to classic risk assessment.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-soa\">\n<dt><a href=\"#dwg-soa\">SoA (Statement of Applicability)<\/a><\/dt>\n<dd>The statement of applicability: which of the controls from Annex A apply \u2013 and, with justification, why individual ones do not.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-annex-a\">\n<dt><a href=\"#dwg-annex-a\">Annex A controls<\/a><\/dt>\n<dd>The catalogue of 38 controls in ISO\/IEC 42001, from which an organisation selects and implements the ones suited to it.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-risk-register\">\n<dt><a href=\"#dwg-risk-register\">Risk register \/ AI risk assessment<\/a><\/dt>\n<dd>The structured recording and treatment of AI risks: likelihood of occurrence, impact, measure and accepted residual risk.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-guardrail\">\n<dt><a href=\"#dwg-guardrail\">Guardrail<\/a><\/dt>\n<dd>A technical or organisational guardrail that keeps an AI system on safe tracks \u2013 e.g. input filters, prompt hardening or tests against prompt injection.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-hitl\">\n<dt><a href=\"#dwg-hitl\">Human-in-the-Loop (HiTL)<\/a><\/dt>\n<dd>The human stays in control: AI outputs are reviewed and approved, not adopted blindly. A core principle of responsible AI.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-bjr\">\n<dt><a href=\"#dwg-bjr\">Business Judgment Rule (BJR)<\/a><\/dt>\n<dd>Liability safe harbour for company directors: a business decision made on an <em>adequate basis of information<\/em>, in good faith and without conflict of interest is protected (\u00a7 93(1) sentence 2 AktG; recognised accordingly for GmbH managing directors). In a dispute, management has to prove that it acted with due care. An AIMS provides and documents the information basis for AI decisions. Whether it was adequate in a given case is a legal question.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-iso-27001\">\n<dt><a href=\"#dwg-iso-27001\">ISO\/IEC 27001 (distinction)<\/a><\/dt>\n<dd>The standard for information security (ISMS) \u2013 not to be confused with ISO\/IEC 42001 (AI). Rule of thumb: security is not the same as AI conformity.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-pdca\">\n<dt><a href=\"#dwg-pdca\">PDCA cycle<\/a><\/dt>\n<dd>Plan \u2013 Do \u2013 Check \u2013 Act: the cycle of continuous improvement on which every management system \u2013 including ISO\/IEC 42001 \u2013 is built.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-internes-audit\">\n<dt><a href=\"#dwg-internes-audit\">Internal audit (Clause 9.2)<\/a><\/dt>\n<dd>The regular, independent check of whether the AIMS works as planned and meets the requirements.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-management-review\">\n<dt><a href=\"#dwg-management-review\">Management review (Clause 9.3)<\/a><\/dt>\n<dd>The periodic assessment of the AIMS by leadership \u2013 the basis for decisions on resources, objectives and improvement.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-nc-capa\">\n<dt><a href=\"#dwg-nc-capa\">NC\/CAPA (Clause 10.2)<\/a><\/dt>\n<dd>Handling non-conformities (Clause 10.2 “Nonconformity and corrective action”): analyse the cause, initiate and document corrective actions. “CAPA” is the common industry shorthand; prevention itself is covered by the standard through risk-based thinking (Clause 6).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-impartiality\">\n<dt><a href=\"#dwg-impartiality\">Impartiality \/ independence<\/a><\/dt>\n<dd>Whoever builds or advises on an AIMS may not also impartially audit or certify it \u2013 build and assessment must stay separate (Clause 9.2; ISO\/IEC 17021).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-reifegrad\">\n<dt><a href=\"#dwg-reifegrad\">Maturity model (maturity check)<\/a><\/dt>\n<dd>An assessment of how far an organisation’s AI governance has developed \u2013 from a first overview to a certification-ready AIMS.<\/dd>\n<\/div>\n<\/dl><\/div>\n<div class=\"dwg-cluster\" data-c=\"tech\">\n<h3>AI technology<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-llm\">\n<dt><a href=\"#dwg-llm\">LLM (Large Language Model)<\/a><\/dt>\n<dd>A large AI language model that understands and generates text \u2013 the “brain” of modern AI assistants.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-token\">\n<dt><a href=\"#dwg-token\">Token<\/a><\/dt>\n<dd>The text chunks a model breaks language into. Performance and cost are calculated in tokens.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-prompt\">\n<dt><a href=\"#dwg-prompt\">Prompt<\/a><\/dt>\n<dd>The instruction or question to the AI. Rule of thumb: quality in = quality out.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-rag\">\n<dt><a href=\"#dwg-rag\">RAG (Retrieval-Augmented Generation)<\/a><\/dt>\n<dd>Before answering, the AI looks things up in stored documents and cites sources \u2013 the basis for reliable knowledge assistants.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-tool-calling\">\n<dt><a href=\"#dwg-tool-calling\">Tool\/Function calling<\/a><\/dt>\n<dd>The AI triggers real actions during the dialogue \u2013 e.g. booking an appointment or calling an interface.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-stt-tts\">\n<dt><a href=\"#dwg-stt-tts\">STT \/ TTS<\/a><\/dt>\n<dd>Speech-to-text (the “ear”: speech \u2192 text) and text-to-speech (the “mouth”: text \u2192 speech) \u2013 the building blocks of a voice agent.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-voice-clone\">\n<dt><a href=\"#dwg-voice-clone\">Voice Clone<\/a><\/dt>\n<dd>An artificially reproduced, recognisable voice. Subject to the AI labelling obligation (Art. 50 AI Act).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-latenz\">\n<dt><a href=\"#dwg-latenz\">Latency<\/a><\/dt>\n<dd>The delay between input and response. On the phone, a conversation feels unnatural beyond ~800 ms.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-model-serving\">\n<dt><a href=\"#dwg-model-serving\">Model Serving<\/a><\/dt>\n<dd>A ready-made service that provides an AI model via an interface \u2013 use it without training your own.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-halluzination\">\n<dt><a href=\"#dwg-halluzination\">Hallucination<\/a><\/dt>\n<dd>When an AI model makes plausible-sounding but false or invented statements \u2013 the reason human review (HiTL) remains necessary.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-fine-tuning\">\n<dt><a href=\"#dwg-fine-tuning\">Fine-Tuning<\/a><\/dt>\n<dd>The targeted re-training of a model with your own data to sharpen it for a specific task.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-embedding\">\n<dt><a href=\"#dwg-embedding\">Embedding \/ vector database<\/a><\/dt>\n<dd>Texts are translated into number vectors and stored in a vector database \u2013 the basis for RAG to find matching sources.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-kontextfenster\">\n<dt><a href=\"#dwg-kontextfenster\">Context window<\/a><\/dt>\n<dd>The amount of text a model can keep “in view” at once. When it is full, older content drops out.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-agent\">\n<dt><a href=\"#dwg-agent\">Agent \/ Agentic AI<\/a><\/dt>\n<dd>An AI that plans and acts autonomously in multiple steps \u2013 using tools, taking intermediate steps, pursuing a goal (e.g. a phone or website agent).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-bias\">\n<dt><a href=\"#dwg-bias\">Bias<\/a><\/dt>\n<dd>Systematic imbalances in data or model that can lead to unfair outcomes \u2013 a core topic of AI fairness and governance.<\/dd>\n<\/div>\n<\/dl><\/div>\n<div class=\"dwg-cluster\" data-c=\"cloud\">\n<h3>Cloud & sovereignty<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-betriebsmodelle\">\n<dt><a href=\"#dwg-betriebsmodelle\">On-Prem \/ IaaS \/ PaaS \/ SaaS<\/a><\/dt>\n<dd>Operating models from “everything yourself” to “just use it”: on-premises (your own house) \u2192 IaaS (rented hardware) \u2192 PaaS (platform) \u2192 SaaS (ready-made software). The closer to SaaS, the less effort \u2013 but the more data at the provider.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-souveraene-cloud\">\n<dt><a href=\"#dwg-souveraene-cloud\">Sovereign cloud<\/a><\/dt>\n<dd>Operation, control and data verifiably located in the EU\/DE, without access by non-European laws (e.g. STACKIT).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-datenhoheit\">\n<dt><a href=\"#dwg-datenhoheit\">Data sovereignty \/ data residency<\/a><\/dt>\n<dd>Sovereignty means: you decide over your data. Residency means: where it physically sits. Both are decisive for governance and GDPR.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-avv\">\n<dt><a href=\"#dwg-avv\">Data processing agreement (Art. 28 GDPR)<\/a><\/dt>\n<dd>The data processing agreement with every service provider in the data path \u2013 mandatory as soon as personal data is processed.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-no-retention\">\n<dt><a href=\"#dwg-no-retention\">No-retention \/ no training<\/a><\/dt>\n<dd>The provider does not store inputs and does not use them for model training \u2013 an important building block of data sovereignty.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-bsi-c5\">\n<dt><a href=\"#dwg-bsi-c5\">BSI C5<\/a><\/dt>\n<dd>A German audit catalogue for cloud security. Important, but: security is not the same as AI Act conformity \u2013 governance comes on top.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-dsgvo\">\n<dt><a href=\"#dwg-dsgvo\">GDPR \/ personal data<\/a><\/dt>\n<dd>The EU General Data Protection Regulation protects personal data. AI governance and data protection are closely intertwined \u2013 e.g. in legal basis, purpose limitation and data-subject rights.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-drittland\">\n<dt><a href=\"#dwg-drittland\">Third-country transfer \/ SCC<\/a><\/dt>\n<dd>If data is processed outside the EU, a suitable basis is needed \u2013 e.g. Standard Contractual Clauses (SCC). Without it, the transfer is legally precarious.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-verschluesselung\">\n<dt><a href=\"#dwg-verschluesselung\">Encryption (at rest \/ in transit)<\/a><\/dt>\n<dd>Data is secured \u2013 in storage (at rest) and in transit. Basic protection for confidentiality and data sovereignty.<\/dd>\n<\/div>\n<\/dl><\/div>\n<p class=\"dwg-empty\">No term found. Try a different spelling?<\/p>\n<div class=\"dwg-foot\">\u00a9 Deepwraite® \u2013 eine Marke der L&L Kommunikationskonzepte GmbH \u00b7 2026 \u00b7 Alle Rechte vorbehalten.<\/div>\n<\/section>\n<p><script type=\"application\/ld+json\">{\"@context\":\"https:\/\/schema.org\",\"@type\":\"DefinedTermSet\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\",\"name\":\"AI Governance Glossary\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/\",\"inLanguage\":\"en\",\"publisher\":{\"@type\":\"Organization\",\"name\":\"Deepwraite\",\"legalName\":\"L&L Kommunikationskonzepte GmbH\"},\"hasDefinedTerm\":[{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-eu-ai-act\",\"name\":\"EU AI Act (AI Regulation)\",\"description\":\"The EU regulation (Reg. 2024\/1689) that regulates AI systems by their risk \u2013 the first comprehensive AI regulation worldwide. It applies in stages and imposes obligations according to risk class.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-eu-ai-act\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risikoklassen\",\"name\":\"Risk classes\",\"description\":\"The AI Act divides AI into four tiers: prohibited (Art. 5), high-risk (via Annex I or III, Art. 6), limited risk (transparency obligation, Art. 50) and minimal risk. The obligations follow from the classification.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risikoklassen\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hochrisiko\",\"name\":\"High-risk AI (Art. 6, Annex I & III)\",\"description\":\"An AI system is high-risk via two routes (Art. 6): the Annex I route \u2013 it is a safety component of a product, or itself a product under the EU harmonisation legislation in Annex I (e.g. machinery, medical devices), and requires third-party conformity assessment; or the Annex III route \u2013 it falls into an area listed there (e.g. employment, education, critical infrastructure, creditworthiness, law enforcement). Consequence: strict obligations such as risk management, technical documentation and human oversight. Exception (Art. 6(3)): Annex III systems without significant risk may be exempt \u2013 but profiling always remains high-risk.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hochrisiko\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anhang-iii\",\"name\":\"Annex III (high-risk areas)\",\"description\":\"The list of stand-alone high-risk applications in the AI Act: biometrics, critical infrastructure, education, employment, access to essential (including private) services \u2013 incl. creditworthiness and insurance \u2013, law enforcement, migration\/border control, and justice and democratic processes.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anhang-iii\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anbieter\",\"name\":\"Provider\",\"description\":\"Whoever develops an AI system and places it on the market under their own name. Carries the main obligations of the AI Act (conformity, documentation, reporting).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anbieter\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betreiber\",\"name\":\"Deployer\",\"description\":\"Whoever uses an AI system under their own responsibility. Must, among other things, ensure human oversight and comply with the provider\u2019s instructions.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betreiber\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-gpai\",\"name\":\"GPAI (General-Purpose AI)\",\"description\":\"General-purpose AI models, such as large language models. Providers of such foundation models are subject to their own transparency and documentation obligations.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-gpai\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-art-50\",\"name\":\"Art. 50 \u2013 Transparency obligation\",\"description\":\"People must be able to recognise that they are interacting with AI or seeing AI-generated content (labelling obligation) \u2013 e.g. an AI announcement at the start of a call or a note on a chatbot.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-art-50\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verbotene-praktiken\",\"name\":\"Prohibited practices (Art. 5)\",\"description\":\"AI applications the AI Act prohibits \u2013 such as social scoring by the state or manipulative systems that deliberately exploit weaknesses.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verbotene-praktiken\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-konformitaetsbewertung\",\"name\":\"Conformity assessment\",\"description\":\"The verification procedure by which \u2013 especially for high-risk AI \u2013 it is demonstrated before market launch that the requirements of the AI Act are met.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-konformitaetsbewertung\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ki-kompetenz\",\"name\":\"AI literacy (Art. 4)\",\"description\":\"Providers and deployers are expected to foster sufficient AI literacy among their staff \u2013 a basic understanding of the opportunities, risks and correct handling of AI. The duty has applied since February 2025. In July 2026 the Digital Omnibus changed the standard from \u201censure\u201d to \u201csupport and promote\u201d: the requirement remains, the degree of obligation is softer. Independently of this, an AIMS under ISO\/IEC 42001 requires competence and awareness in clauses 7.2 and 7.3 \u2013 there the evidence is still expected.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ki-kompetenz\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fria\",\"name\":\"Fundamental-rights impact assessment (FRIA, Art. 27)\",\"description\":\"Certain deployers of high-risk AI must assess the impact on the fundamental rights of affected persons before use (Fundamental Rights Impact Assessment).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fria\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-post-market\",\"name\":\"Post-market monitoring (Art. 72)\",\"description\":\"Providers of high-risk AI must actively monitor their system after placing it on the market and systematically evaluate findings.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-post-market\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-vorfallmeldung\",\"name\":\"Serious incidents (Art. 73)\",\"description\":\"Providers must report serious incidents and malfunctions of their AI to the authorities \u2013 within defined deadlines.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-vorfallmeldung\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ce\",\"name\":\"CE marking & declaration of conformity\",\"description\":\"With the EU declaration of conformity and CE marking, the provider confirms that a high-risk AI system meets the requirements and may be placed on the market.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ce\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-42001\",\"name\":\"ISO\/IEC 42001 (AIMS)\",\"description\":\"The international standard for an AI management system: it defines how an organisation governs, documents and continuously improves AI responsibly. Certifiable \u2013 Deepwraite\u00ae\u2019s core field.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-42001\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aims\",\"name\":\"AIMS (AI Management System)\",\"description\":\"The living management system for AI: roles, policies, risks, controls and evidence across an organisation\u2019s entire use of AI \u2013 not a document, but a process.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aims\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kik\",\"name\":\"AI coordinator (KIK)\",\"description\":\"The central operational role in the AIMS: coordinates set-up and operation, maintains the documentation, manages risks and approvals (ISO\/IEC 42001, Clause 5.3).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kik\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aisia\",\"name\":\"AISIA (AI System Impact Assessment)\",\"description\":\"The assessment of an AI system\u2019s impact on people, groups and fundamental rights (ISO\/IEC 42001, Clause 6.1.4) \u2013 the AI-specific complement to classic risk assessment.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aisia\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-soa\",\"name\":\"SoA (Statement of Applicability)\",\"description\":\"The statement of applicability: which of the controls from Annex A apply \u2013 and, with justification, why individual ones do not.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-soa\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-annex-a\",\"name\":\"Annex A controls\",\"description\":\"The catalogue of 38 controls in ISO\/IEC 42001, from which an organisation selects and implements the ones suited to it.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-annex-a\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risk-register\",\"name\":\"Risk register \/ AI risk assessment\",\"description\":\"The structured recording and treatment of AI risks: likelihood of occurrence, impact, measure and accepted residual risk.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risk-register\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-guardrail\",\"name\":\"Guardrail\",\"description\":\"A technical or organisational guardrail that keeps an AI system on safe tracks \u2013 e.g. input filters, prompt hardening or tests against prompt injection.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-guardrail\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hitl\",\"name\":\"Human-in-the-Loop (HiTL)\",\"description\":\"The human stays in control: AI outputs are reviewed and approved, not adopted blindly. A core principle of responsible AI.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hitl\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bjr\",\"name\":\"Business Judgment Rule (BJR)\",\"description\":\"Liability safe harbour for company directors: a business decision made on an adequate basis of information, in good faith and without conflict of interest is protected (\u00a7 93(1) sentence 2 AktG; recognised accordingly for GmbH managing directors). In a dispute, management has to prove that it acted with due care. An AIMS provides and documents the information basis for AI decisions. Whether it was adequate in a given case is a legal question.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bjr\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-27001\",\"name\":\"ISO\/IEC 27001 (distinction)\",\"description\":\"The standard for information security (ISMS) \u2013 not to be confused with ISO\/IEC 42001 (AI). Rule of thumb: security is not the same as AI conformity.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-27001\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-pdca\",\"name\":\"PDCA cycle\",\"description\":\"Plan \u2013 Do \u2013 Check \u2013 Act: the cycle of continuous improvement on which every management system \u2013 including ISO\/IEC 42001 \u2013 is built.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-pdca\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-internes-audit\",\"name\":\"Internal audit (Clause 9.2)\",\"description\":\"The regular, independent check of whether the AIMS works as planned and meets the requirements.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-internes-audit\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-management-review\",\"name\":\"Management review (Clause 9.3)\",\"description\":\"The periodic assessment of the AIMS by leadership \u2013 the basis for decisions on resources, objectives and improvement.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-management-review\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-nc-capa\",\"name\":\"NC\/CAPA (Clause 10.2)\",\"description\":\"Handling non-conformities (Clause 10.2 \u201cNonconformity and corrective action\u201d): analyse the cause, initiate and document corrective actions. \u201cCAPA\u201d is the common industry shorthand; prevention itself is covered by the standard through risk-based thinking (Clause 6).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-nc-capa\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-impartiality\",\"name\":\"Impartiality \/ independence\",\"description\":\"Whoever builds or advises on an AIMS may not also impartially audit or certify it \u2013 build and assessment must stay separate (Clause 9.2; ISO\/IEC 17021).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-impartiality\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-reifegrad\",\"name\":\"Maturity model (maturity check)\",\"description\":\"An assessment of how far an organisation\u2019s AI governance has developed \u2013 from a first overview to a certification-ready AIMS.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-reifegrad\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-llm\",\"name\":\"LLM (Large Language Model)\",\"description\":\"A large AI language model that understands and generates text \u2013 the \u201cbrain\u201d of modern AI assistants.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-llm\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-token\",\"name\":\"Token\",\"description\":\"The text chunks a model breaks language into. Performance and cost are calculated in tokens.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-token\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-prompt\",\"name\":\"Prompt\",\"description\":\"The instruction or question to the AI. Rule of thumb: quality in = quality out.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-prompt\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-rag\",\"name\":\"RAG (Retrieval-Augmented Generation)\",\"description\":\"Before answering, the AI looks things up in stored documents and cites sources \u2013 the basis for reliable knowledge assistants.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-rag\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-tool-calling\",\"name\":\"Tool\/Function calling\",\"description\":\"The AI triggers real actions during the dialogue \u2013 e.g. booking an appointment or calling an interface.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-tool-calling\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-stt-tts\",\"name\":\"STT \/ TTS\",\"description\":\"Speech-to-text (the \u201cear\u201d: speech \u2192 text) and text-to-speech (the \u201cmouth\u201d: text \u2192 speech) \u2013 the building blocks of a voice agent.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-stt-tts\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-voice-clone\",\"name\":\"Voice Clone\",\"description\":\"An artificially reproduced, recognisable voice. Subject to the AI labelling obligation (Art. 50 AI Act).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-voice-clone\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-latenz\",\"name\":\"Latency\",\"description\":\"The delay between input and response. On the phone, a conversation feels unnatural beyond ~800 ms.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-latenz\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-model-serving\",\"name\":\"Model Serving\",\"description\":\"A ready-made service that provides an AI model via an interface \u2013 use it without training your own.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-model-serving\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-halluzination\",\"name\":\"Hallucination\",\"description\":\"When an AI model makes plausible-sounding but false or invented statements \u2013 the reason human review (HiTL) remains necessary.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-halluzination\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fine-tuning\",\"name\":\"Fine-Tuning\",\"description\":\"The targeted re-training of a model with your own data to sharpen it for a specific task.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fine-tuning\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-embedding\",\"name\":\"Embedding \/ vector database\",\"description\":\"Texts are translated into number vectors and stored in a vector database \u2013 the basis for RAG to find matching sources.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-embedding\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kontextfenster\",\"name\":\"Context window\",\"description\":\"The amount of text a model can keep \u201cin view\u201d at once. When it is full, older content drops out.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kontextfenster\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-agent\",\"name\":\"Agent \/ Agentic AI\",\"description\":\"An AI that plans and acts autonomously in multiple steps \u2013 using tools, taking intermediate steps, pursuing a goal (e.g. a phone or website agent).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-agent\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bias\",\"name\":\"Bias\",\"description\":\"Systematic imbalances in data or model that can lead to unfair outcomes \u2013 a core topic of AI fairness and governance.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bias\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betriebsmodelle\",\"name\":\"On-Prem \/ IaaS \/ PaaS \/ SaaS\",\"description\":\"Operating models from \u201ceverything yourself\u201d to \u201cjust use it\u201d: on-premises (your own house) \u2192 IaaS (rented hardware) \u2192 PaaS (platform) \u2192 SaaS (ready-made software). The closer to SaaS, the less effort \u2013 but the more data at the provider.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betriebsmodelle\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-souveraene-cloud\",\"name\":\"Sovereign cloud\",\"description\":\"Operation, control and data verifiably located in the EU\/DE, without access by non-European laws (e.g. STACKIT).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-souveraene-cloud\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-datenhoheit\",\"name\":\"Data sovereignty \/ data residency\",\"description\":\"Sovereignty means: you decide over your data. Residency means: where it physically sits. Both are decisive for governance and GDPR.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-datenhoheit\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-avv\",\"name\":\"Data processing agreement (Art. 28 GDPR)\",\"description\":\"The data processing agreement with every service provider in the data path \u2013 mandatory as soon as personal data is processed.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-avv\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-no-retention\",\"name\":\"No-retention \/ no training\",\"description\":\"The provider does not store inputs and does not use them for model training \u2013 an important building block of data sovereignty.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-no-retention\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bsi-c5\",\"name\":\"BSI C5\",\"description\":\"A German audit catalogue for cloud security. Important, but: security is not the same as AI Act conformity \u2013 governance comes on top.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bsi-c5\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-dsgvo\",\"name\":\"GDPR \/ personal data\",\"description\":\"The EU General Data Protection Regulation protects personal data. AI governance and data protection are closely intertwined \u2013 e.g. in legal basis, purpose limitation and data-subject rights.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-dsgvo\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-drittland\",\"name\":\"Third-country transfer \/ SCC\",\"description\":\"If data is processed outside the EU, a suitable basis is needed \u2013 e.g. Standard Contractual Clauses (SCC). Without it, the transfer is legally precarious.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-drittland\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verschluesselung\",\"name\":\"Encryption (at rest \/ in transit)\",\"description\":\"Data is secured \u2013 in storage (at rest) and in transit. Basic protection for confidentiality and data sovereignty.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verschluesselung\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}}]}<\/script><\/p>\n<p><script>\n(function(){\n var root=document.currentScript.parentNode.querySelector('.dwg-glossar')||document.querySelector('.dwg-glossar');\n if(!root)return;\n var terms=[].slice.call(root.querySelectorAll('.dwg-term'));\n var q=root.querySelector('.dwg-q'), chips=root.querySelector('.dwg-chips'),\n count=root.querySelector('.dwg-count'), empty=root.querySelector('.dwg-empty'),\n sections=[].slice.call(root.querySelectorAll('.dwg-cluster')), filter='all';\n terms.forEach(function(t){t._a=t.querySelector('dt a');t._dd=t.querySelector('dd');\n t._aTxt=t._a.textContent;t._ddHTML=t._dd.innerHTML;t._hay=(t._a.textContent+' '+t._dd.textContent).toLowerCase();});\n function esc(s){return s.replace(\/[.*+?^${}()|[\\]\\\\]\/g,'\\\\$&');}\n function hlDD(orig,rx){return rx?orig.replace(\/>([^<]+)<\/g,function(m,x){return '>'+x.replace(rx,'<mark>$1<\/mark>')+'<';}).replace(\/^([^<]+)\/,function(m,x){return x.replace(rx,'<mark>$1<\/mark>');}):orig;}\n function run(){\n var term=q.value.trim().toLowerCase(),rx=term?new RegExp('('+esc(term)+')','ig'):null,shown=0;\n terms.forEach(function(t){\n var inC=(filter==='all')||(t.closest('.dwg-cluster').getAttribute('data-c')===filter);\n var match=inC&&(!term||t._hay.indexOf(term)>-1);\n t.style.display=match?'':'none';\n if(match){shown++;t._a.innerHTML=rx?t._aTxt.replace(rx,'<mark>$1<\/mark>'):t._aTxt;t._dd.innerHTML=hlDD(t._ddHTML,rx);}\n });\n sections.forEach(function(s){\n var vis=[].some.call(s.querySelectorAll('.dwg-term'),function(t){return t.style.display!=='none';});\n s.style.display=(vis&&(filter==='all'||s.getAttribute('data-c')===filter))?'':'none';\n });\n empty.style.display=shown?'none':'block';\n count.textContent=shown+' terms';\n }\n q.addEventListener('input',run);\n chips.addEventListener('click',function(e){var c=e.target.closest('.dwg-chip');if(!c)return;\n filter=c.getAttribute('data-f');\n [].forEach.call(chips.children,function(x){x.setAttribute('aria-pressed',x===c?'true':'false');});run();});\n run();\n})();\n<\/script><br \/>\n<\/p>\n<p><\/body><br \/>\n<\/html><\/p>\n\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>AI Governance Glossary (Embed) \u2013 Deepwraite AI governance knowledge Governance Glossary The key terms around introducing an AI management system based on ISO\/IEC 42001. All EU AI Act ISO 42001 & governance AI technology Cloud & sovereignty EU AI Act EU AI Act (AI Regulation) The EU regulation (Reg. 2024\/1689) that regulates AI systems by their […]<\/p>\n","protected":false},"author":2,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-2336","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages\/2336","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/comments?post=2336"}],"version-history":[{"count":21,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages\/2336\/revisions"}],"predecessor-version":[{"id":3301,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages\/2336\/revisions\/3301"}],"wp:attachment":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/media?parent=2336"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

\n\t\t\t\t

\n\t\t\t\t\t

\n\t\t\t\t

\n\t\t\t\t\t\t\t\t\t\n\n

\n
\n
\n
\n
\nAI Governance Glossary (Embed) \u2013 Deepwraite<\/title><br \/>\n<\/p>\n<style>body{margin:0;background:#0b0e11}<\/style>\n<p><\/head><br \/>\n<body data-rsssl=1><\/p>\n<p><\/p>\n<style>\n.dwg-glossar{--dwg-teal:#0ee2cf;--dwg-teal-deep:#0a8f83;--dwg-card:rgba(255,255,255,.035);\n --dwg-line:rgba(255,255,255,.10);--dwg-ink:#e9eef3;--dwg-muted:#8b97a3;\n font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;\n color:var(--dwg-ink);box-sizing:border-box;width:100%;max-width:1340px;margin:28px auto;position:relative;isolation:isolate;\n border:1px solid rgba(255,255,255,.12);border-radius:16px;\n background:radial-gradient(120% 120% at 88% 8%,rgba(14,226,207,.10),transparent 48%),#111110;\n padding:clamp(26px,4vw,40px) 28px;}\n.dwg-glossar::before{content:\"\";position:absolute;inset:0;z-index:-1;border-radius:16px;\n background-image:radial-gradient(#FFFFFF0D 1px,transparent 1px);background-size:22px 22px;\n -webkit-mask-image:radial-gradient(120% 80% at 50% 0%,#000 25%,transparent 72%);\n mask-image:radial-gradient(120% 80% at 50% 0%,#000 25%,transparent 72%);}\n.dwg-glossar *{box-sizing:border-box}\n.dwg-glossar .dwg-brand{margin:0 0 10px;font-weight:800;letter-spacing:.06em;color:var(--dwg-teal);font-size:12.5px;text-transform:uppercase}\n.dwg-glossar .dwg-h1{font-size:30px;line-height:1.18;margin:0 0 12px;font-weight:800;letter-spacing:-.01em;color:#fff}\n.dwg-glossar .dwg-lead{color:var(--dwg-muted);max-width:70ch;margin:0 0 4px;font-size:16px;line-height:1.6}\n.dwg-glossar .dwg-controls{position:sticky;top:0;padding:16px 0 12px;margin:24px 0 6px;\n background:linear-gradient(#111110 72%,rgba(17,17,16,0));z-index:4;border-bottom:1px solid var(--dwg-line)}\n.dwg-glossar .dwg-q{width:100%;padding:13px 15px;border:1px solid var(--dwg-line);border-radius:12px;font:inherit;\n font-size:16px;background:rgba(255,255,255,.04);color:var(--dwg-ink);outline:none}\n.dwg-glossar .dwg-q::placeholder{color:#6a7681}\n.dwg-glossar .dwg-q:focus{border-color:var(--dwg-teal-deep);box-shadow:0 0 0 3px rgba(14,226,207,.18)}\n.dwg-glossar .dwg-chips{display:flex;flex-wrap:wrap;gap:8px;margin-top:12px}\n.dwg-glossar .dwg-chip{border:1px solid var(--dwg-line);background:rgba(255,255,255,.03);color:var(--dwg-muted);\n padding:7px 14px;border-radius:999px;font-size:13px;font-weight:600;cursor:pointer;user-select:none;transition:.15s}\n.dwg-glossar .dwg-chip:hover{border-color:var(--dwg-teal-deep);color:var(--dwg-ink)}\n.dwg-glossar .dwg-chip[aria-pressed=\"true\"]{background:var(--dwg-teal);border-color:var(--dwg-teal);color:#08211f}\n.dwg-glossar .dwg-count{color:var(--dwg-muted);font-size:13px;margin:2px 2px 20px}\n.dwg-glossar .dwg-cluster{margin:0 0 34px}\n.dwg-glossar .dwg-cluster>h3{font-size:19px;margin:0 0 14px;padding-left:12px;border-left:3px solid var(--dwg-teal);\n letter-spacing:-.01em;color:#fff;font-weight:700}\n.dwg-glossar dl{margin:0;display:grid;grid-template-columns:1fr;gap:12px}\n.dwg-glossar .dwg-term{background:var(--dwg-card);border:1px solid var(--dwg-line);border-radius:14px;\n padding:16px 18px;scroll-margin-top:120px;transition:border-color .15s}\n.dwg-glossar .dwg-term:hover{border-color:rgba(14,226,207,.35)}\n.dwg-glossar .dwg-term dt{font-weight:700;font-size:16.5px;margin:0 0 5px}\n.dwg-glossar .dwg-term dt a{color:#fff;text-decoration:none}\n.dwg-glossar .dwg-term dt a:hover{color:var(--dwg-teal)}\n.dwg-glossar .dwg-term dd{margin:0;color:#c3ccd4;font-size:15px}\n.dwg-glossar mark{background:rgba(14,226,207,.30);color:#fff;border-radius:3px;padding:0 1px}\n.dwg-glossar .dwg-empty{display:none;color:var(--dwg-muted);padding:30px 4px}\n.dwg-glossar .dwg-foot{border-top:1px solid var(--dwg-line);margin-top:38px;padding-top:16px;\n color:var(--dwg-muted);font-size:12px;text-align:center}\n@media(min-width:720px){.dwg-glossar dl{grid-template-columns:1fr 1fr}.dwg-glossar .dwg-h1{font-size:37px}}\n<\/style>\n<section class=\"dwg-glossar\" role=\"region\" aria-label=\"AI Governance Glossary\">\n<p class=\"dwg-brand\">AI governance knowledge<\/p>\n<h2 class=\"dwg-h1\">Governance Glossary<\/h2>\n<p class=\"dwg-lead\">The key terms around introducing an AI management system based on <strong>ISO\/IEC 42001<\/strong>.<\/p>\n<div class=\"dwg-controls\">\n <input class=\"dwg-q\" type=\"search\" placeholder=\"Search term \u2026 (e.g. AIMS, high-risk, token, sovereign)\" aria-label=\"Search glossary\"><\/p>\n<div class=\"dwg-chips\">\n <span class=\"dwg-chip\" data-f=\"all\" aria-pressed=\"true\">All<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"aiact\" aria-pressed=\"false\">EU AI Act<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"iso\" aria-pressed=\"false\">ISO 42001 & governance<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"tech\" aria-pressed=\"false\">AI technology<\/span><br \/>\n <span class=\"dwg-chip\" data-f=\"cloud\" aria-pressed=\"false\">Cloud & sovereignty<\/span>\n <\/div>\n<\/p><\/div>\n<div class=\"dwg-count\"><\/div>\n<div class=\"dwg-cluster\" data-c=\"aiact\">\n<h3>EU AI Act<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-eu-ai-act\">\n<dt><a href=\"#dwg-eu-ai-act\">EU AI Act (AI Regulation)<\/a><\/dt>\n<dd>The EU regulation (Reg. 2024\/1689) that regulates AI systems by their risk \u2013 the first comprehensive AI regulation worldwide. It applies in stages and imposes obligations according to risk class.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-risikoklassen\">\n<dt><a href=\"#dwg-risikoklassen\">Risk classes<\/a><\/dt>\n<dd>The AI Act divides AI into four tiers: <em>prohibited<\/em> (Art. 5), <em>high-risk<\/em> (via Annex I or III, Art. 6), <em>limited risk<\/em> (transparency obligation, Art. 50) and <em>minimal risk<\/em>. The obligations follow from the classification.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-hochrisiko\">\n<dt><a href=\"#dwg-hochrisiko\">High-risk AI (Art. 6, Annex I & III)<\/a><\/dt>\n<dd>An AI system is high-risk via <strong>two routes<\/strong> (Art. 6): the <strong>Annex I route<\/strong> \u2013 it is a safety component of a product, or itself a product under the EU harmonisation legislation in Annex I (e.g. machinery, medical devices), and requires third-party conformity assessment; or the <strong>Annex III route<\/strong> \u2013 it falls into an area listed there (e.g. employment, education, critical infrastructure, creditworthiness, law enforcement). Consequence: strict obligations such as risk management, technical documentation and human oversight. Exception (Art. 6(3)): Annex III systems without significant risk may be exempt \u2013 but <em>profiling<\/em> always remains high-risk.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-anhang-iii\">\n<dt><a href=\"#dwg-anhang-iii\">Annex III (high-risk areas)<\/a><\/dt>\n<dd>The list of stand-alone high-risk applications in the AI Act: biometrics, critical infrastructure, education, employment, access to essential (including private) services \u2013 incl. creditworthiness and insurance \u2013, law enforcement, migration\/border control, and justice and democratic processes.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-anbieter\">\n<dt><a href=\"#dwg-anbieter\">Provider<\/a><\/dt>\n<dd>Whoever develops an AI system and places it on the market under their own name. Carries the main obligations of the AI Act (conformity, documentation, reporting).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-betreiber\">\n<dt><a href=\"#dwg-betreiber\">Deployer<\/a><\/dt>\n<dd>Whoever uses an AI system under their own responsibility. Must, among other things, ensure human oversight and comply with the provider’s instructions.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-gpai\">\n<dt><a href=\"#dwg-gpai\">GPAI (General-Purpose AI)<\/a><\/dt>\n<dd>General-purpose AI models, such as large language models. Providers of such foundation models are subject to their own transparency and documentation obligations.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-art-50\">\n<dt><a href=\"#dwg-art-50\">Art. 50 \u2013 Transparency obligation<\/a><\/dt>\n<dd>People must be able to recognise that they are interacting with AI or seeing AI-generated content (labelling obligation) \u2013 e.g. an AI announcement at the start of a call or a note on a chatbot.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-verbotene-praktiken\">\n<dt><a href=\"#dwg-verbotene-praktiken\">Prohibited practices (Art. 5)<\/a><\/dt>\n<dd>AI applications the AI Act prohibits \u2013 such as social scoring by the state or manipulative systems that deliberately exploit weaknesses.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-konformitaetsbewertung\">\n<dt><a href=\"#dwg-konformitaetsbewertung\">Conformity assessment<\/a><\/dt>\n<dd>The verification procedure by which \u2013 especially for high-risk AI \u2013 it is demonstrated before market launch that the requirements of the AI Act are met.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-ki-kompetenz\">\n<dt><a href=\"#dwg-ki-kompetenz\">AI literacy (Art. 4)<\/a><\/dt>\n<dd>Providers and deployers are expected to foster sufficient AI literacy among their staff \u2013 a basic understanding of the opportunities, risks and correct handling of AI. The duty has applied since February 2025. In July 2026 the Digital Omnibus changed the standard from \u201censure\u201d to \u201csupport and promote\u201d: the requirement remains, the degree of obligation is softer. Independently of this, an AIMS under ISO\/IEC 42001 requires competence and awareness in clauses 7.2 and 7.3 \u2013 there the evidence is still expected.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-fria\">\n<dt><a href=\"#dwg-fria\">Fundamental-rights impact assessment (FRIA, Art. 27)<\/a><\/dt>\n<dd>Certain deployers of high-risk AI must assess the impact on the fundamental rights of affected persons before use (Fundamental Rights Impact Assessment).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-post-market\">\n<dt><a href=\"#dwg-post-market\">Post-market monitoring (Art. 72)<\/a><\/dt>\n<dd>Providers of high-risk AI must actively monitor their system after placing it on the market and systematically evaluate findings.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-vorfallmeldung\">\n<dt><a href=\"#dwg-vorfallmeldung\">Serious incidents (Art. 73)<\/a><\/dt>\n<dd>Providers must report serious incidents and malfunctions of their AI to the authorities \u2013 within defined deadlines.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-ce\">\n<dt><a href=\"#dwg-ce\">CE marking & declaration of conformity<\/a><\/dt>\n<dd>With the EU declaration of conformity and CE marking, the provider confirms that a high-risk AI system meets the requirements and may be placed on the market.<\/dd>\n<\/div>\n<\/dl><\/div>\n<div class=\"dwg-cluster\" data-c=\"iso\">\n<h3>ISO\/IEC 42001 & governance<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-iso-42001\">\n<dt><a href=\"#dwg-iso-42001\">ISO\/IEC 42001 (AIMS)<\/a><\/dt>\n<dd>The international standard for an AI management system: it defines how an organisation governs, documents and continuously improves AI responsibly. Certifiable \u2013 Deepwraite\u00ae’s core field.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-aims\">\n<dt><a href=\"#dwg-aims\">AIMS (AI Management System)<\/a><\/dt>\n<dd>The living management system for AI: roles, policies, risks, controls and evidence across an organisation’s entire use of AI \u2013 not a document, but a process.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-kik\">\n<dt><a href=\"#dwg-kik\">AI coordinator (KIK)<\/a><\/dt>\n<dd>The central operational role in the AIMS: coordinates set-up and operation, maintains the documentation, manages risks and approvals (ISO\/IEC 42001, Clause 5.3).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-aisia\">\n<dt><a href=\"#dwg-aisia\">AISIA (AI System Impact Assessment)<\/a><\/dt>\n<dd>The assessment of an AI system’s impact on people, groups and fundamental rights (ISO\/IEC 42001, Clause 6.1.4) \u2013 the AI-specific complement to classic risk assessment.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-soa\">\n<dt><a href=\"#dwg-soa\">SoA (Statement of Applicability)<\/a><\/dt>\n<dd>The statement of applicability: which of the controls from Annex A apply \u2013 and, with justification, why individual ones do not.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-annex-a\">\n<dt><a href=\"#dwg-annex-a\">Annex A controls<\/a><\/dt>\n<dd>The catalogue of 38 controls in ISO\/IEC 42001, from which an organisation selects and implements the ones suited to it.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-risk-register\">\n<dt><a href=\"#dwg-risk-register\">Risk register \/ AI risk assessment<\/a><\/dt>\n<dd>The structured recording and treatment of AI risks: likelihood of occurrence, impact, measure and accepted residual risk.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-guardrail\">\n<dt><a href=\"#dwg-guardrail\">Guardrail<\/a><\/dt>\n<dd>A technical or organisational guardrail that keeps an AI system on safe tracks \u2013 e.g. input filters, prompt hardening or tests against prompt injection.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-hitl\">\n<dt><a href=\"#dwg-hitl\">Human-in-the-Loop (HiTL)<\/a><\/dt>\n<dd>The human stays in control: AI outputs are reviewed and approved, not adopted blindly. A core principle of responsible AI.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-bjr\">\n<dt><a href=\"#dwg-bjr\">Business Judgment Rule (BJR)<\/a><\/dt>\n<dd>Liability safe harbour for company directors: a business decision made on an <em>adequate basis of information<\/em>, in good faith and without conflict of interest is protected (\u00a7 93(1) sentence 2 AktG; recognised accordingly for GmbH managing directors). In a dispute, management has to prove that it acted with due care. An AIMS provides and documents the information basis for AI decisions. Whether it was adequate in a given case is a legal question.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-iso-27001\">\n<dt><a href=\"#dwg-iso-27001\">ISO\/IEC 27001 (distinction)<\/a><\/dt>\n<dd>The standard for information security (ISMS) \u2013 not to be confused with ISO\/IEC 42001 (AI). Rule of thumb: security is not the same as AI conformity.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-pdca\">\n<dt><a href=\"#dwg-pdca\">PDCA cycle<\/a><\/dt>\n<dd>Plan \u2013 Do \u2013 Check \u2013 Act: the cycle of continuous improvement on which every management system \u2013 including ISO\/IEC 42001 \u2013 is built.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-internes-audit\">\n<dt><a href=\"#dwg-internes-audit\">Internal audit (Clause 9.2)<\/a><\/dt>\n<dd>The regular, independent check of whether the AIMS works as planned and meets the requirements.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-management-review\">\n<dt><a href=\"#dwg-management-review\">Management review (Clause 9.3)<\/a><\/dt>\n<dd>The periodic assessment of the AIMS by leadership \u2013 the basis for decisions on resources, objectives and improvement.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-nc-capa\">\n<dt><a href=\"#dwg-nc-capa\">NC\/CAPA (Clause 10.2)<\/a><\/dt>\n<dd>Handling non-conformities (Clause 10.2 “Nonconformity and corrective action”): analyse the cause, initiate and document corrective actions. “CAPA” is the common industry shorthand; prevention itself is covered by the standard through risk-based thinking (Clause 6).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-impartiality\">\n<dt><a href=\"#dwg-impartiality\">Impartiality \/ independence<\/a><\/dt>\n<dd>Whoever builds or advises on an AIMS may not also impartially audit or certify it \u2013 build and assessment must stay separate (Clause 9.2; ISO\/IEC 17021).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-reifegrad\">\n<dt><a href=\"#dwg-reifegrad\">Maturity model (maturity check)<\/a><\/dt>\n<dd>An assessment of how far an organisation’s AI governance has developed \u2013 from a first overview to a certification-ready AIMS.<\/dd>\n<\/div>\n<\/dl><\/div>\n<div class=\"dwg-cluster\" data-c=\"tech\">\n<h3>AI technology<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-llm\">\n<dt><a href=\"#dwg-llm\">LLM (Large Language Model)<\/a><\/dt>\n<dd>A large AI language model that understands and generates text \u2013 the “brain” of modern AI assistants.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-token\">\n<dt><a href=\"#dwg-token\">Token<\/a><\/dt>\n<dd>The text chunks a model breaks language into. Performance and cost are calculated in tokens.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-prompt\">\n<dt><a href=\"#dwg-prompt\">Prompt<\/a><\/dt>\n<dd>The instruction or question to the AI. Rule of thumb: quality in = quality out.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-rag\">\n<dt><a href=\"#dwg-rag\">RAG (Retrieval-Augmented Generation)<\/a><\/dt>\n<dd>Before answering, the AI looks things up in stored documents and cites sources \u2013 the basis for reliable knowledge assistants.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-tool-calling\">\n<dt><a href=\"#dwg-tool-calling\">Tool\/Function calling<\/a><\/dt>\n<dd>The AI triggers real actions during the dialogue \u2013 e.g. booking an appointment or calling an interface.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-stt-tts\">\n<dt><a href=\"#dwg-stt-tts\">STT \/ TTS<\/a><\/dt>\n<dd>Speech-to-text (the “ear”: speech \u2192 text) and text-to-speech (the “mouth”: text \u2192 speech) \u2013 the building blocks of a voice agent.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-voice-clone\">\n<dt><a href=\"#dwg-voice-clone\">Voice Clone<\/a><\/dt>\n<dd>An artificially reproduced, recognisable voice. Subject to the AI labelling obligation (Art. 50 AI Act).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-latenz\">\n<dt><a href=\"#dwg-latenz\">Latency<\/a><\/dt>\n<dd>The delay between input and response. On the phone, a conversation feels unnatural beyond ~800 ms.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-model-serving\">\n<dt><a href=\"#dwg-model-serving\">Model Serving<\/a><\/dt>\n<dd>A ready-made service that provides an AI model via an interface \u2013 use it without training your own.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-halluzination\">\n<dt><a href=\"#dwg-halluzination\">Hallucination<\/a><\/dt>\n<dd>When an AI model makes plausible-sounding but false or invented statements \u2013 the reason human review (HiTL) remains necessary.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-fine-tuning\">\n<dt><a href=\"#dwg-fine-tuning\">Fine-Tuning<\/a><\/dt>\n<dd>The targeted re-training of a model with your own data to sharpen it for a specific task.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-embedding\">\n<dt><a href=\"#dwg-embedding\">Embedding \/ vector database<\/a><\/dt>\n<dd>Texts are translated into number vectors and stored in a vector database \u2013 the basis for RAG to find matching sources.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-kontextfenster\">\n<dt><a href=\"#dwg-kontextfenster\">Context window<\/a><\/dt>\n<dd>The amount of text a model can keep “in view” at once. When it is full, older content drops out.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-agent\">\n<dt><a href=\"#dwg-agent\">Agent \/ Agentic AI<\/a><\/dt>\n<dd>An AI that plans and acts autonomously in multiple steps \u2013 using tools, taking intermediate steps, pursuing a goal (e.g. a phone or website agent).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-bias\">\n<dt><a href=\"#dwg-bias\">Bias<\/a><\/dt>\n<dd>Systematic imbalances in data or model that can lead to unfair outcomes \u2013 a core topic of AI fairness and governance.<\/dd>\n<\/div>\n<\/dl><\/div>\n<div class=\"dwg-cluster\" data-c=\"cloud\">\n<h3>Cloud & sovereignty<\/h3>\n<dl>\n<div class=\"dwg-term\" id=\"dwg-betriebsmodelle\">\n<dt><a href=\"#dwg-betriebsmodelle\">On-Prem \/ IaaS \/ PaaS \/ SaaS<\/a><\/dt>\n<dd>Operating models from “everything yourself” to “just use it”: on-premises (your own house) \u2192 IaaS (rented hardware) \u2192 PaaS (platform) \u2192 SaaS (ready-made software). The closer to SaaS, the less effort \u2013 but the more data at the provider.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-souveraene-cloud\">\n<dt><a href=\"#dwg-souveraene-cloud\">Sovereign cloud<\/a><\/dt>\n<dd>Operation, control and data verifiably located in the EU\/DE, without access by non-European laws (e.g. STACKIT).<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-datenhoheit\">\n<dt><a href=\"#dwg-datenhoheit\">Data sovereignty \/ data residency<\/a><\/dt>\n<dd>Sovereignty means: you decide over your data. Residency means: where it physically sits. Both are decisive for governance and GDPR.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-avv\">\n<dt><a href=\"#dwg-avv\">Data processing agreement (Art. 28 GDPR)<\/a><\/dt>\n<dd>The data processing agreement with every service provider in the data path \u2013 mandatory as soon as personal data is processed.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-no-retention\">\n<dt><a href=\"#dwg-no-retention\">No-retention \/ no training<\/a><\/dt>\n<dd>The provider does not store inputs and does not use them for model training \u2013 an important building block of data sovereignty.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-bsi-c5\">\n<dt><a href=\"#dwg-bsi-c5\">BSI C5<\/a><\/dt>\n<dd>A German audit catalogue for cloud security. Important, but: security is not the same as AI Act conformity \u2013 governance comes on top.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-dsgvo\">\n<dt><a href=\"#dwg-dsgvo\">GDPR \/ personal data<\/a><\/dt>\n<dd>The EU General Data Protection Regulation protects personal data. AI governance and data protection are closely intertwined \u2013 e.g. in legal basis, purpose limitation and data-subject rights.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-drittland\">\n<dt><a href=\"#dwg-drittland\">Third-country transfer \/ SCC<\/a><\/dt>\n<dd>If data is processed outside the EU, a suitable basis is needed \u2013 e.g. Standard Contractual Clauses (SCC). Without it, the transfer is legally precarious.<\/dd>\n<\/div>\n<div class=\"dwg-term\" id=\"dwg-verschluesselung\">\n<dt><a href=\"#dwg-verschluesselung\">Encryption (at rest \/ in transit)<\/a><\/dt>\n<dd>Data is secured \u2013 in storage (at rest) and in transit. Basic protection for confidentiality and data sovereignty.<\/dd>\n<\/div>\n<\/dl><\/div>\n<p class=\"dwg-empty\">No term found. Try a different spelling?<\/p>\n<div class=\"dwg-foot\">\u00a9 Deepwraite® \u2013 eine Marke der L&L Kommunikationskonzepte GmbH \u00b7 2026 \u00b7 Alle Rechte vorbehalten.<\/div>\n<\/section>\n<p><script type=\"application\/ld+json\">{\"@context\":\"https:\/\/schema.org\",\"@type\":\"DefinedTermSet\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\",\"name\":\"AI Governance Glossary\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/\",\"inLanguage\":\"en\",\"publisher\":{\"@type\":\"Organization\",\"name\":\"Deepwraite\",\"legalName\":\"L&L Kommunikationskonzepte GmbH\"},\"hasDefinedTerm\":[{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-eu-ai-act\",\"name\":\"EU AI Act (AI Regulation)\",\"description\":\"The EU regulation (Reg. 2024\/1689) that regulates AI systems by their risk \u2013 the first comprehensive AI regulation worldwide. It applies in stages and imposes obligations according to risk class.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-eu-ai-act\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risikoklassen\",\"name\":\"Risk classes\",\"description\":\"The AI Act divides AI into four tiers: prohibited (Art. 5), high-risk (via Annex I or III, Art. 6), limited risk (transparency obligation, Art. 50) and minimal risk. The obligations follow from the classification.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risikoklassen\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hochrisiko\",\"name\":\"High-risk AI (Art. 6, Annex I & III)\",\"description\":\"An AI system is high-risk via two routes (Art. 6): the Annex I route \u2013 it is a safety component of a product, or itself a product under the EU harmonisation legislation in Annex I (e.g. machinery, medical devices), and requires third-party conformity assessment; or the Annex III route \u2013 it falls into an area listed there (e.g. employment, education, critical infrastructure, creditworthiness, law enforcement). Consequence: strict obligations such as risk management, technical documentation and human oversight. Exception (Art. 6(3)): Annex III systems without significant risk may be exempt \u2013 but profiling always remains high-risk.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hochrisiko\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anhang-iii\",\"name\":\"Annex III (high-risk areas)\",\"description\":\"The list of stand-alone high-risk applications in the AI Act: biometrics, critical infrastructure, education, employment, access to essential (including private) services \u2013 incl. creditworthiness and insurance \u2013, law enforcement, migration\/border control, and justice and democratic processes.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anhang-iii\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anbieter\",\"name\":\"Provider\",\"description\":\"Whoever develops an AI system and places it on the market under their own name. Carries the main obligations of the AI Act (conformity, documentation, reporting).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-anbieter\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betreiber\",\"name\":\"Deployer\",\"description\":\"Whoever uses an AI system under their own responsibility. Must, among other things, ensure human oversight and comply with the provider\u2019s instructions.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betreiber\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-gpai\",\"name\":\"GPAI (General-Purpose AI)\",\"description\":\"General-purpose AI models, such as large language models. Providers of such foundation models are subject to their own transparency and documentation obligations.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-gpai\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-art-50\",\"name\":\"Art. 50 \u2013 Transparency obligation\",\"description\":\"People must be able to recognise that they are interacting with AI or seeing AI-generated content (labelling obligation) \u2013 e.g. an AI announcement at the start of a call or a note on a chatbot.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-art-50\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verbotene-praktiken\",\"name\":\"Prohibited practices (Art. 5)\",\"description\":\"AI applications the AI Act prohibits \u2013 such as social scoring by the state or manipulative systems that deliberately exploit weaknesses.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verbotene-praktiken\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-konformitaetsbewertung\",\"name\":\"Conformity assessment\",\"description\":\"The verification procedure by which \u2013 especially for high-risk AI \u2013 it is demonstrated before market launch that the requirements of the AI Act are met.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-konformitaetsbewertung\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ki-kompetenz\",\"name\":\"AI literacy (Art. 4)\",\"description\":\"Providers and deployers are expected to foster sufficient AI literacy among their staff \u2013 a basic understanding of the opportunities, risks and correct handling of AI. The duty has applied since February 2025. In July 2026 the Digital Omnibus changed the standard from \u201censure\u201d to \u201csupport and promote\u201d: the requirement remains, the degree of obligation is softer. Independently of this, an AIMS under ISO\/IEC 42001 requires competence and awareness in clauses 7.2 and 7.3 \u2013 there the evidence is still expected.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ki-kompetenz\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fria\",\"name\":\"Fundamental-rights impact assessment (FRIA, Art. 27)\",\"description\":\"Certain deployers of high-risk AI must assess the impact on the fundamental rights of affected persons before use (Fundamental Rights Impact Assessment).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fria\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-post-market\",\"name\":\"Post-market monitoring (Art. 72)\",\"description\":\"Providers of high-risk AI must actively monitor their system after placing it on the market and systematically evaluate findings.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-post-market\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-vorfallmeldung\",\"name\":\"Serious incidents (Art. 73)\",\"description\":\"Providers must report serious incidents and malfunctions of their AI to the authorities \u2013 within defined deadlines.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-vorfallmeldung\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ce\",\"name\":\"CE marking & declaration of conformity\",\"description\":\"With the EU declaration of conformity and CE marking, the provider confirms that a high-risk AI system meets the requirements and may be placed on the market.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-ce\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-42001\",\"name\":\"ISO\/IEC 42001 (AIMS)\",\"description\":\"The international standard for an AI management system: it defines how an organisation governs, documents and continuously improves AI responsibly. Certifiable \u2013 Deepwraite\u00ae\u2019s core field.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-42001\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aims\",\"name\":\"AIMS (AI Management System)\",\"description\":\"The living management system for AI: roles, policies, risks, controls and evidence across an organisation\u2019s entire use of AI \u2013 not a document, but a process.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aims\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kik\",\"name\":\"AI coordinator (KIK)\",\"description\":\"The central operational role in the AIMS: coordinates set-up and operation, maintains the documentation, manages risks and approvals (ISO\/IEC 42001, Clause 5.3).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kik\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aisia\",\"name\":\"AISIA (AI System Impact Assessment)\",\"description\":\"The assessment of an AI system\u2019s impact on people, groups and fundamental rights (ISO\/IEC 42001, Clause 6.1.4) \u2013 the AI-specific complement to classic risk assessment.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-aisia\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-soa\",\"name\":\"SoA (Statement of Applicability)\",\"description\":\"The statement of applicability: which of the controls from Annex A apply \u2013 and, with justification, why individual ones do not.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-soa\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-annex-a\",\"name\":\"Annex A controls\",\"description\":\"The catalogue of 38 controls in ISO\/IEC 42001, from which an organisation selects and implements the ones suited to it.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-annex-a\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risk-register\",\"name\":\"Risk register \/ AI risk assessment\",\"description\":\"The structured recording and treatment of AI risks: likelihood of occurrence, impact, measure and accepted residual risk.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-risk-register\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-guardrail\",\"name\":\"Guardrail\",\"description\":\"A technical or organisational guardrail that keeps an AI system on safe tracks \u2013 e.g. input filters, prompt hardening or tests against prompt injection.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-guardrail\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hitl\",\"name\":\"Human-in-the-Loop (HiTL)\",\"description\":\"The human stays in control: AI outputs are reviewed and approved, not adopted blindly. A core principle of responsible AI.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-hitl\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bjr\",\"name\":\"Business Judgment Rule (BJR)\",\"description\":\"Liability safe harbour for company directors: a business decision made on an adequate basis of information, in good faith and without conflict of interest is protected (\u00a7 93(1) sentence 2 AktG; recognised accordingly for GmbH managing directors). In a dispute, management has to prove that it acted with due care. An AIMS provides and documents the information basis for AI decisions. Whether it was adequate in a given case is a legal question.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bjr\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-27001\",\"name\":\"ISO\/IEC 27001 (distinction)\",\"description\":\"The standard for information security (ISMS) \u2013 not to be confused with ISO\/IEC 42001 (AI). Rule of thumb: security is not the same as AI conformity.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-iso-27001\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-pdca\",\"name\":\"PDCA cycle\",\"description\":\"Plan \u2013 Do \u2013 Check \u2013 Act: the cycle of continuous improvement on which every management system \u2013 including ISO\/IEC 42001 \u2013 is built.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-pdca\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-internes-audit\",\"name\":\"Internal audit (Clause 9.2)\",\"description\":\"The regular, independent check of whether the AIMS works as planned and meets the requirements.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-internes-audit\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-management-review\",\"name\":\"Management review (Clause 9.3)\",\"description\":\"The periodic assessment of the AIMS by leadership \u2013 the basis for decisions on resources, objectives and improvement.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-management-review\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-nc-capa\",\"name\":\"NC\/CAPA (Clause 10.2)\",\"description\":\"Handling non-conformities (Clause 10.2 \u201cNonconformity and corrective action\u201d): analyse the cause, initiate and document corrective actions. \u201cCAPA\u201d is the common industry shorthand; prevention itself is covered by the standard through risk-based thinking (Clause 6).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-nc-capa\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-impartiality\",\"name\":\"Impartiality \/ independence\",\"description\":\"Whoever builds or advises on an AIMS may not also impartially audit or certify it \u2013 build and assessment must stay separate (Clause 9.2; ISO\/IEC 17021).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-impartiality\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-reifegrad\",\"name\":\"Maturity model (maturity check)\",\"description\":\"An assessment of how far an organisation\u2019s AI governance has developed \u2013 from a first overview to a certification-ready AIMS.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-reifegrad\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-llm\",\"name\":\"LLM (Large Language Model)\",\"description\":\"A large AI language model that understands and generates text \u2013 the \u201cbrain\u201d of modern AI assistants.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-llm\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-token\",\"name\":\"Token\",\"description\":\"The text chunks a model breaks language into. Performance and cost are calculated in tokens.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-token\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-prompt\",\"name\":\"Prompt\",\"description\":\"The instruction or question to the AI. Rule of thumb: quality in = quality out.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-prompt\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-rag\",\"name\":\"RAG (Retrieval-Augmented Generation)\",\"description\":\"Before answering, the AI looks things up in stored documents and cites sources \u2013 the basis for reliable knowledge assistants.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-rag\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-tool-calling\",\"name\":\"Tool\/Function calling\",\"description\":\"The AI triggers real actions during the dialogue \u2013 e.g. booking an appointment or calling an interface.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-tool-calling\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-stt-tts\",\"name\":\"STT \/ TTS\",\"description\":\"Speech-to-text (the \u201cear\u201d: speech \u2192 text) and text-to-speech (the \u201cmouth\u201d: text \u2192 speech) \u2013 the building blocks of a voice agent.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-stt-tts\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-voice-clone\",\"name\":\"Voice Clone\",\"description\":\"An artificially reproduced, recognisable voice. Subject to the AI labelling obligation (Art. 50 AI Act).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-voice-clone\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-latenz\",\"name\":\"Latency\",\"description\":\"The delay between input and response. On the phone, a conversation feels unnatural beyond ~800 ms.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-latenz\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-model-serving\",\"name\":\"Model Serving\",\"description\":\"A ready-made service that provides an AI model via an interface \u2013 use it without training your own.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-model-serving\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-halluzination\",\"name\":\"Hallucination\",\"description\":\"When an AI model makes plausible-sounding but false or invented statements \u2013 the reason human review (HiTL) remains necessary.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-halluzination\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fine-tuning\",\"name\":\"Fine-Tuning\",\"description\":\"The targeted re-training of a model with your own data to sharpen it for a specific task.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-fine-tuning\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-embedding\",\"name\":\"Embedding \/ vector database\",\"description\":\"Texts are translated into number vectors and stored in a vector database \u2013 the basis for RAG to find matching sources.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-embedding\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kontextfenster\",\"name\":\"Context window\",\"description\":\"The amount of text a model can keep \u201cin view\u201d at once. When it is full, older content drops out.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-kontextfenster\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-agent\",\"name\":\"Agent \/ Agentic AI\",\"description\":\"An AI that plans and acts autonomously in multiple steps \u2013 using tools, taking intermediate steps, pursuing a goal (e.g. a phone or website agent).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-agent\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bias\",\"name\":\"Bias\",\"description\":\"Systematic imbalances in data or model that can lead to unfair outcomes \u2013 a core topic of AI fairness and governance.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bias\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betriebsmodelle\",\"name\":\"On-Prem \/ IaaS \/ PaaS \/ SaaS\",\"description\":\"Operating models from \u201ceverything yourself\u201d to \u201cjust use it\u201d: on-premises (your own house) \u2192 IaaS (rented hardware) \u2192 PaaS (platform) \u2192 SaaS (ready-made software). The closer to SaaS, the less effort \u2013 but the more data at the provider.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-betriebsmodelle\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-souveraene-cloud\",\"name\":\"Sovereign cloud\",\"description\":\"Operation, control and data verifiably located in the EU\/DE, without access by non-European laws (e.g. STACKIT).\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-souveraene-cloud\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-datenhoheit\",\"name\":\"Data sovereignty \/ data residency\",\"description\":\"Sovereignty means: you decide over your data. Residency means: where it physically sits. Both are decisive for governance and GDPR.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-datenhoheit\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-avv\",\"name\":\"Data processing agreement (Art. 28 GDPR)\",\"description\":\"The data processing agreement with every service provider in the data path \u2013 mandatory as soon as personal data is processed.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-avv\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-no-retention\",\"name\":\"No-retention \/ no training\",\"description\":\"The provider does not store inputs and does not use them for model training \u2013 an important building block of data sovereignty.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-no-retention\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bsi-c5\",\"name\":\"BSI C5\",\"description\":\"A German audit catalogue for cloud security. Important, but: security is not the same as AI Act conformity \u2013 governance comes on top.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-bsi-c5\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-dsgvo\",\"name\":\"GDPR \/ personal data\",\"description\":\"The EU General Data Protection Regulation protects personal data. AI governance and data protection are closely intertwined \u2013 e.g. in legal basis, purpose limitation and data-subject rights.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-dsgvo\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-drittland\",\"name\":\"Third-country transfer \/ SCC\",\"description\":\"If data is processed outside the EU, a suitable basis is needed \u2013 e.g. Standard Contractual Clauses (SCC). Without it, the transfer is legally precarious.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-drittland\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}},{\"@type\":\"DefinedTerm\",\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verschluesselung\",\"name\":\"Encryption (at rest \/ in transit)\",\"description\":\"Data is secured \u2013 in storage (at rest) and in transit. Basic protection for confidentiality and data sovereignty.\",\"url\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#dwg-verschluesselung\",\"inDefinedTermSet\":{\"@id\":\"https:\/\/www.deepwraite.de\/en\/governance-glossary\/#glossary\"}}]}<\/script><\/p>\n<p><script>\n(function(){\n var root=document.currentScript.parentNode.querySelector('.dwg-glossar')||document.querySelector('.dwg-glossar');\n if(!root)return;\n var terms=[].slice.call(root.querySelectorAll('.dwg-term'));\n var q=root.querySelector('.dwg-q'), chips=root.querySelector('.dwg-chips'),\n count=root.querySelector('.dwg-count'), empty=root.querySelector('.dwg-empty'),\n sections=[].slice.call(root.querySelectorAll('.dwg-cluster')), filter='all';\n terms.forEach(function(t){t._a=t.querySelector('dt a');t._dd=t.querySelector('dd');\n t._aTxt=t._a.textContent;t._ddHTML=t._dd.innerHTML;t._hay=(t._a.textContent+' '+t._dd.textContent).toLowerCase();});\n function esc(s){return s.replace(\/[.*+?^${}()|[\\]\\\\]\/g,'\\\\$&');}\n function hlDD(orig,rx){return rx?orig.replace(\/>([^<]+)<\/g,function(m,x){return '>'+x.replace(rx,'<mark>$1<\/mark>')+'<';}).replace(\/^([^<]+)\/,function(m,x){return x.replace(rx,'<mark>$1<\/mark>');}):orig;}\n function run(){\n var term=q.value.trim().toLowerCase(),rx=term?new RegExp('('+esc(term)+')','ig'):null,shown=0;\n terms.forEach(function(t){\n var inC=(filter==='all')||(t.closest('.dwg-cluster').getAttribute('data-c')===filter);\n var match=inC&&(!term||t._hay.indexOf(term)>-1);\n t.style.display=match?'':'none';\n if(match){shown++;t._a.innerHTML=rx?t._aTxt.replace(rx,'<mark>$1<\/mark>'):t._aTxt;t._dd.innerHTML=hlDD(t._ddHTML,rx);}\n });\n sections.forEach(function(s){\n var vis=[].some.call(s.querySelectorAll('.dwg-term'),function(t){return t.style.display!=='none';});\n s.style.display=(vis&&(filter==='all'||s.getAttribute('data-c')===filter))?'':'none';\n });\n empty.style.display=shown?'none':'block';\n count.textContent=shown+' terms';\n }\n q.addEventListener('input',run);\n chips.addEventListener('click',function(e){var c=e.target.closest('.dwg-chip');if(!c)return;\n filter=c.getAttribute('data-f');\n [].forEach.call(chips.children,function(x){x.setAttribute('aria-pressed',x===c?'true':'false');});run();});\n run();\n})();\n<\/script><br \/>\n<\/p>\n<p><\/body><br \/>\n<\/html><\/p>\n\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>AI Governance Glossary (Embed) \u2013 Deepwraite AI governance knowledge Governance Glossary The key terms around introducing an AI management system based on ISO\/IEC 42001. All EU AI Act ISO 42001 & governance AI technology Cloud & sovereignty EU AI Act EU AI Act (AI Regulation) The EU regulation (Reg. 2024\/1689) that regulates AI systems by their […]<\/p>\n","protected":false},"author":2,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-2336","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages\/2336","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/comments?post=2336"}],"version-history":[{"count":21,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages\/2336\/revisions"}],"predecessor-version":[{"id":3301,"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/pages\/2336\/revisions\/3301"}],"wp:attachment":[{"href":"https:\/\/www.deepwraite.de\/en\/wp-json\/wp\/v2\/media?parent=2336"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}