The register
Explore all deployer risks here.
Risks are organized into 7 families corresponding to enterprise risk domains. Families are built from the MIT AI Risk Repository and enriched with MITRE ATLAS, ISO, and the EU AI Act.
7families
82canonical risks
61MITRE ATLAS sub-risks
10frameworks crosswalked
| Family | Risks | Sub-risks | Enterprise risk domain | MITRE ATLAS | ISO | EU AI Act |
|---|---|---|---|---|---|---|
Model & system behaviour
How the AI system itself behaves: bias, toxic or unsafe output, hallucination, brittleness, and emergent capability. |
29 | 5 | Operational & technology risk | 5 sub-risks | 29 refs | 7 refs |
Governance & process
Accountability, oversight, documentation, evaluation, and the lifecycle discipline of running AI. |
13 | 0 | Operational & governance risk | not decomposed | +7 gaps | 8 refs |
Regulatory compliance
Duties under the EU AI Act and sector rules: impact assessments, registration, notice, and monitoring. |
12 | 0 | Compliance & legal risk | not decomposed | 1 ref | +11 gaps |
Human & usage
How people interact with, rely on, or misuse the system: manipulation, overreliance, and loss of human agency. |
10 | 0 | Conduct & operational risk | not decomposed | 10 refs | 4 refs |
Security & adversarial
Attacks on the AI system: prompt injection, evasion, poisoning, model theft, and autonomous-agent abuse. |
7 | 43 | Cyber & information security risk | 43 sub-risks | 6 refs | 1 ref |
Data, privacy & content liability
Personal-data exposure, unlawful processing, confidentiality, and content-related legal liability. |
6 | 5 | Privacy, data & legal risk | 5 sub-risks | 6 refs | 2 refs |
Third party & supply chain
Risks inherited from model providers, vendors, and the AI supply chain: concentration, version churn, embedded AI. |
5 | 8 | Third-party & supply-chain risk | 8 sub-risks | +2 gaps | 2 refs |
N sub-risks decomposed by MITRE ATLAS
+N gaps risks the source added
N refs cross-referenced only
Enterprise risk domains match the family band on the home page. ISO clause and EU article references cross-cut families; the badges mark where a source introduced or decomposed risks. Family colours carry through to the full register below.
Browse
The full register.
82 canonical risks; the 12 risks with MITRE ATLAS-anchored sub-risks are badged. Filtering runs in the browser.
82 of 82 risks
| Id | Risk | Family | AI type | Scope | Source |
|---|---|---|---|---|---|
| MR-001 | Biased or discriminatory outputs and decisions | Model & system behaviour | GPAI, Agentic, Classical_ML | Both | MIT AI Risk Repository v4 |
| MR-002 | Stereotyping and representational harm | Model & system behaviour | GPAI, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-003 | Toxic, hateful, or harassing content generation | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-004 | Violent or extremist content generation | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-005 | Child sexual abuse material and child-safety harm | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-006 | Disparate performance across groups and languages | Model & system behaviour | GPAI, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-007 | Self-harm, suicide and dangerous-behavior promotion | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-008 | Sexual content and non-consensual intimate imagery | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-009 | Leakage of personal or sensitive data3 sub-risks | Data, privacy & content liability | GPAI, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-010 | Prompt injection and jailbreaking10 sub-risks | Security & adversarial | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-011 | Unlawful or non-consensual collection and processing of personal data | Data, privacy & content liability | GPAI, Classical_ML | Both | MIT AI Risk Repository v4 |
| MR-012 | Adversarial examples and evasion attacks4 sub-risks | Security & adversarial | Classical_ML, GPAI | System | MIT AI Risk Repository v4 |
| MR-013 | Disclosure of confidential or proprietary information2 sub-risks | Data, privacy & content liability | GPAI, Agentic | Both | MIT AI Risk Repository v4 |
| MR-014 | Data and model poisoning and backdoors4 sub-risks | Security & adversarial | GPAI, Classical_ML, Agentic | System | MIT AI Risk Repository v4 |
| MR-015 | Residual AI system security and availability weaknesses9 sub-risks | Security & adversarial | GPAI, Classical_ML, Agentic | System | MIT AI Risk Repository v4 |
| MR-016 | Model theft, extraction and weight leakage3 sub-risks | Security & adversarial | GPAI, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-017 | Privacy-invasive inference and re-identification | Data, privacy & content liability | GPAI, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-018 | AI supply-chain and infrastructure vulnerabilities8 sub-risks | Third party & supply chain | GPAI, Classical_ML, Agentic | Both | MIT AI Risk Repository v4 |
| MR-019 | Insecure or vulnerable code generation | Model & system behaviour | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-020 | Insecure integration with external tools, plugins and APIs | Security & adversarial | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-021 | Hallucination and fabricated output | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-022 | Unsafe or incorrect advice in high-stakes domains | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-023 | Contribution to misinformation and information-ecosystem degradation | Model & system behaviour | GPAI | Both | MIT AI Risk Repository v4 |
| MR-024 | Defamation and false statements about people | Data, privacy & content liability | GPAI | System | MIT AI Risk Repository v4 |
| MR-025 | Overconfidence, sycophancy and poor calibration | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-026 | Disinformation and influence operations | Model & system behaviour | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-027 | AI-enabled cyberattacks and offensive cyber operations1 sub-risks | Model & system behaviour | GPAI, Agentic | Both | MIT AI Risk Repository v4 |
| MR-028 | AI-enabled fraud, scams and social engineering2 sub-risks | Model & system behaviour | GPAI, Agentic | Both | MIT AI Risk Repository v4 |
| MR-029 | Facilitation of weapons, CBRN and serious physical harm (capability uplift) | Model & system behaviour | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-030 | Manipulation, persuasion and dark patterns | Human & usage | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-031 | Impersonation, deepfakes and synthetic media2 sub-risks | Model & system behaviour | GPAI | Both | MIT AI Risk Repository v4 |
| MR-032 | Deliberate misuse and repurposing for harm | Human & usage | GPAI, Agentic | Both | MIT AI Risk Repository v4 |
| MR-033 | Mass surveillance and censorship enablement | Human & usage | GPAI, Classical_ML | Both | MIT AI Risk Repository v4 |
| MR-034 | Overreliance and automation bias | Human & usage | GPAI, Agentic, Classical_ML | Both | MIT AI Risk Repository v4 |
| MR-035 | Erosion of human agency and autonomy | Human & usage | GPAI, Agentic | Both | MIT AI Risk Repository v4 |
| MR-036 | Anthropomorphism, emotional dependence and psychological harm | Human & usage | GPAI | System | MIT AI Risk Repository v4 |
| MR-037 | Environmental footprint of AI | Governance & process | GPAI, Classical_ML, Agentic | Both | MIT AI Risk Repository v4 |
| MR-038 | Workforce displacement and job-quality decline | Human & usage | GPAI, Agentic, Classical_ML | Organization | MIT AI Risk Repository v4 |
| MR-039 | Intellectual property and copyright infringement | Data, privacy & content liability | GPAI | Both | MIT AI Risk Repository v4 |
| MR-040 | Regulatory non-compliance and legal liability | Regulatory compliance | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-041 | Academic and professional dishonesty | Human & usage | GPAI | System | MIT AI Risk Repository v4 |
| MR-042 | Unclear accountability and responsibility for AI decisions | Governance & process | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-043 | Inadequate AI governance and oversight processes | Governance & process | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-044 | Exploitative labor in the AI supply chain | Third party & supply chain | GPAI, Classical_ML | Organization | MIT AI Risk Repository v4 |
| MR-045 | Insufficient documentation, transparency and data provenance | Governance & process | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-046 | Inadequate evaluation, testing and benchmarking | Governance & process | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-047 | Vendor/model concentration, monoculture and correlated failure | Third party & supply chain | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-048 | AI competence and skills gaps in the organization | Human & usage | GPAI, Classical_ML, Agentic | Organization | MIT AI Risk Repository v4 |
| MR-049 | Physical safety harm and accidents | Model & system behaviour | GPAI, Agentic, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-050 | Inaccuracy and poor predictive performance | Model & system behaviour | GPAI, Classical_ML, Agentic | System | MIT AI Risk Repository v4 |
| MR-051 | Ethical or value misalignment in outputs and decisions | Model & system behaviour | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-052 | Emergent dangerous capabilities | Model & system behaviour | GPAI, Agentic | System | MIT AI Risk Repository v4 |
| MR-053 | Goal misspecification, specification gaming and reward hacking | Model & system behaviour | GPAI, Agentic, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-054 | Loss of meaningful human oversight and control | Governance & process | Agentic, GPAI | Both | MIT AI Risk Repository v4 |
| MR-055 | Lack of explainability and interpretability | Model & system behaviour | GPAI, Classical_ML, Agentic | Both | MIT AI Risk Repository v4 |
| MR-056 | Lack of robustness to distribution shift and edge cases | Model & system behaviour | GPAI, Classical_ML, Agentic | System | MIT AI Risk Repository v4 |
| MR-057 | Multi-agent interaction risks | Model & system behaviour | Agentic, GPAI | Both | MIT AI Risk Repository v4 |
| MR-058 | Model performance drift and degradation over time | Model & system behaviour | GPAI, Classical_ML | System | MIT AI Risk Repository v4 |
| MR-059 | Poor data quality and representativeness | Model & system behaviour | GPAI, Classical_ML | Both | MIT AI Risk Repository v4 |
| MR-060 | Use outside intended scope | Human & usage | GPAI, Agentic, Classical_ML | Both | MIT AI Risk Repository v4 |
| MR-061 | Over-refusal and excessive safety filtering | Model & system behaviour | GPAI | System | MIT AI Risk Repository v4 |
| MR-062 | Shadow AI and unsanctioned use of AI tools | Governance & process | GPAI, Agentic, Classical_ML | Organization | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-063 | AI inventory blind spots | Governance & process | GPAI, Agentic, Classical_ML | Organization | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-064 | Embedded AI introduced through procurement | Third party & supply chain | GPAI, Classical_ML | Organization | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-065 | Vendor model version churn and undisclosed updates | Third party & supply chain | GPAI, Agentic | Both | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-066 | Change deployed without revalidation | Governance & process | GPAI, Classical_ML, Agentic | System | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-067 | Absence of AI impact assessment | Governance & process | GPAI, Agentic, Classical_ML | Organization | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-068 | Inadequate AI incident response and communication | Governance & process | GPAI, Agentic, Classical_ML | Organization | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-069 | Inadequate logging, record-keeping and traceability | Governance & process | GPAI, Agentic, Classical_ML | System | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-070 | Inadequate AI decommissioning and retirement | Governance & process | GPAI, Classical_ML, Agentic | Organization | ISO/IEC 23894 + 42001 (gap analysis) |
| MR-071 | Autonomous agent hijacking and excessive-agency abuse13 sub-risks | Security & adversarial | Agentic, GPAI | Both | MITRE ATLAS v5.6.0 (gap analysis) |
| MR-072 | Failure to conduct a fundamental rights impact assessment | Regulatory compliance | GPAI, Agentic, Classical_ML | Organization | EU AI Act 2024/1689 (gap analysis) |
| MR-073 | Failure to inform workers before workplace deployment | Regulatory compliance | GPAI, Agentic, Classical_ML | Organization | EU AI Act 2024/1689 (gap analysis) |
| MR-074 | Failure to register a high-risk system or verify its registration | Regulatory compliance | GPAI, Agentic, Classical_ML | Organization | EU AI Act 2024/1689 (gap analysis) |
| MR-075 | Failure to inform individuals subject to high-risk AI decisions | Regulatory compliance | GPAI, Agentic, Classical_ML | Both | EU AI Act 2024/1689 (gap analysis) |
| MR-076 | Failure to monitor operation and meet incident-reporting and suspension duties | Regulatory compliance | GPAI, Agentic, Classical_ML | Organization | EU AI Act 2024/1689 (gap analysis) |
| MR-077 | Deploying or repurposing a system into a prohibited practice | Regulatory compliance | GPAI, Agentic, Classical_ML | Both | EU AI Act 2024/1689 (gap analysis) |
| MR-078 | Failure to meet AI transparency and disclosure obligations | Regulatory compliance | GPAI, Agentic | Both | EU AI Act 2024/1689 (gap analysis) |
| MR-079 | Non-compliance with the AI literacy obligation | Regulatory compliance | GPAI, Agentic, Classical_ML | Organization | EU AI Act 2024/1689 (gap analysis) |
| MR-080 | Provider fails to supply adequate GPAI model documentation | Regulatory compliance | GPAI, Agentic | Both | GPAI Code of Practice 2025 (gap analysis) |
| MR-081 | Provider's inadequate copyright compliance exposes the deployer to IP liability | Regulatory compliance | GPAI, Agentic | Both | GPAI Code of Practice 2025 (gap analysis) |
| MR-082 | Provider's inadequate systemic-risk safety and security management, with no deployer visibility | Regulatory compliance | GPAI, Agentic | Both | GPAI Code of Practice 2025 (gap analysis) |
Part of the Deployer AI Risk Register, an open-source resource developed by MindXO. Version 1.0, 3 July 2026. Derived from the MIT AI Risk Repository (V4, December 2025) under CC BY 4.0; an independent derivative work, not endorsed by or affiliated with MIT. Sub-risk decomposition references MITRE ATLAS™ v5.6.0 (© 2021-2026 The MITRE Corporation, reproduced and distributed with permission). ISO/IEC and EU AI Act references are by number only. License: CC BY 4.0. Full attribution and licensing.