AI Incidents and Hazards Monitor
OECD · Framework · OECD
High impact. A reviewer confirms each change before it is listed here.
Our summary
The OECD operates a monitoring system that tracks actual AI incidents and hazards reported in international news media to build an evidence base for AI policy discussions. The system uses news aggregation and machine learning to automatically identify and classify events where AI systems cause or could cause documented harm across multiple categories including health, critical infrastructure, human rights, and environmental damage. Events are enriched with metadata and grouped daily to provide policymakers with real-time insights into global AI-related risks.
Written by the Index in our own words. It is not legal advice; the official text is the authority.
Inside this policy
Every section of this policy in order. Our notes summarise each one in plain language and do not quote the official text, which is the authority.
Drafted by the Index An automated summary of the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
6 sections
Jump to a chapter
Overview
The OECD.AI expert group developed the AI Incidents and Hazards Monitor to track actual AI incidents and hazards in real time from reputable international news outlets processed by Event Registry, which analyzes over 150,000 news articles daily. An open submission process and complementary sources such as court rulings and supervisory authority decisions are being developed to build a more complete evidence base. Data collection and analysis are conducted to ensure reliability, objectivity and quality of information.
Definitions
Information Transparency Disclosures
Methodology for monitoring AI incidents and hazards
2 sections
Since November 2024, events tagged with AI-related concepts are retrieved from Event Registry and classified into incidents, hazards or unrelated events using large language models in a two-step process: GPT-4o mini filters unrelated events, then GPT-4o reclassifies remaining events. Individual articles are classified using the same process based on full article content rather than summaries, and if articles within an event differ, the event adopts the most common classification. Historical incidents and hazards collected before November 2024 are being retrospectively annotated using the same methodology.
Update frequency
The system processes events from one to four days old daily, optimising captured information while allowing media coverage time. News items concerning ongoing incidents or hazards are collected from Event Registry for four days following the event and added to the corresponding record, capturing coverage during peak media attention; articles after this period are likely miscategorised as separate events.
Summary table and process map of the methodology
AI events and their associated articles are identified and categorised as incidents, hazards or unrelated using OECD definitions; approximately 1,000 AI events are analysed daily yielding roughly 10 incidents or hazards. Metadata including title, country, industry and affected stakeholders are added, and articles are appended to events up to four days after initial registration; roughly 30 incidents and hazards are detected daily overall. Before publication, new records are checked and grouped with existing events as appropriate.
Controls
What this policy requires, forbids, grants or penalises, one control at a time and in our own words. Each control links to the official text, which is the authority. This is not legal advice.
Drafted by the Index Drafted automatically from the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
Themes covered 0 of 35
Choose a theme to see only its controls. About the themes
Risk and scope 0 of 4
- Risk classification, not covered
- Prohibited practices, not covered
- Risk management system, not covered
- Impact assessment, not covered
Data and privacy 0 of 4
- Data governance and quality, not covered
- Data subject rights and consent, not covered
- Cross-border data transfer, not covered
- Copyright and training data, not covered
Transparency and records 0 of 4
- Transparency and disclosure to users, not covered
- Labelling of synthetic content, not covered
- Technical documentation, not covered
- Record keeping and logging, not covered
Technical safeguards 0 of 4
- Human oversight, not covered
- Accuracy, robustness and cybersecurity, not covered
- Bias and non-discrimination, not covered
- Testing and evaluation, not covered
Assurance and lifecycle 0 of 4
- Conformity assessment and certification, not covered
- Registration and notification, not covered
- Incident and breach reporting, not covered
- Post-market monitoring and corrective action, not covered
Rights and protected groups 0 of 4
- Automated decision-making rights, not covered
- Children and vulnerable groups, not covered
- Biometric identification and surveillance, not covered
- Whistleblowing, complaints and redress, not covered
Roles and organisation 0 of 6
- Supply chain and provider-deployer duties, not covered
- General purpose and foundation models, not covered
- Accountability and governance roles, not covered
- Training and AI literacy, not covered
- Public sector use and procurement, not covered
- Innovation support and sandboxes, not covered
Enforcement and sectors 0 of 5
- Enforcement, penalties and regulators, not covered
- Sector rules: health, not covered
- Sector rules: finance and insurance, not covered
- Sector rules: employment, not covered
- Sector rules: education, not covered
The full list of controls could not be loaded, so only the first 0 are shown.
Change history
No changes recorded since we began monitoring this policy.