| Canonical name | MIT AI Risk Repository |
| Address | https://airisk.mit.edu/risks |
| Operator | MIT AI Risk Initiative (MIT FutureTech) op |
| Kind | database |
| Roles | harm |
| Lifecycle | active (version 4, December 2025) |
| Observation | 2026-10-28: read |
| Unit | risk, extracted from published risk frameworks op |
| Count | more than 1,700 risks from 74 frameworks (version 4, December 2025); the Initiative's home page gives 65+ (as of 2026-10-29) op |
| Licence | CC BY 4.0 ('unless a specific page states otherwise') op |
| Note | itself an aggregator of frameworks; real-world incidents are kept separately in its AI Incident Tracker |
op operator statement · 3p third-party report · obs our observation · rd our reading
Its unit is the risk as a framework states it. Each source framework is summarized, and the source documents and key excerpts are published with it. The risks are sorted into a domain taxonomy with subdomains; the April 2025 update added a subdomain for multi-agent risks.
It aggregates aggregators: its sources are frameworks, which are themselves syntheses. What enters is what a framework named. A risk no framework has written down has no row.
Its incident layer is a second-order reading of the AI Incident Database, done by machine. The repository sits over the harm basin as a taxonomy for it.
Lee Sharks, ed. "MIT AI Risk Repository." MMRS meta-aggregator, Machine-Mediated Reception Studies, compiled 2026-10-29. https://www.machinemediation.org/aggregators/mit-ai-risk-repository/
Record https://www.machinemediation.org/aggregators/mit-ai-risk-repository/ · entity https://www.machinemediation.org/aggregators/mit-ai-risk-repository/#entity