AI Incidents — AI incidents, model failures, and adversarial-use cases — dated and sourced.
Deepfake Fraud Cases Explained: What 8 Incidents Show
Deepfake fraud cases explained: Arup's HK$200M video call, the WPP and Ferrari near-misses, FBI and FinCEN warnings, and what stops these scams.
Latest reports
Start here
The catalog, the methodology behind it, and the guides analysts reach for most: what to look up, where an incident is recorded, what a regulator must be told, and how to run the response.
- AI Incident Explorer
65 curated incidents and 16 disclosed vulnerabilities, filterable by type, harm domain, modality, actor class and source tier. Exports APA and BibTeX.
- Incident-response playbook
The standard lifecycle with the AI-specific detection sources, containment levers and evidence types added.
- Who you have to notify
EU AI Act Article 73, GDPR, SEC Item 1.05 and sector regulators, with the deadline attached to each trigger.
- Which incident database to use
AIID, OECD AIM, AIAAIC, AVID and MITRE ATLAS record different things. What each one holds, side by side.
- Deepfake detection tools
Reality Defender, Intel FakeCatcher, Pindrop Pulse, Sensity AI and C2PA provenance, compared by threat surface and architecture.
- Reading an ML-library CVE
NVD entries for torch, transformers, vLLM and LangChain, field by field, and what to do when the metadata is thin.
Independent, specialist, and free to read
AI Incidents publishes focused, sourced guides on a single topic. No paywall, no account, no ad tracking.
AI Incidents — in your inbox
AI incidents, model failures, and adversarial-use cases — dated and sourced — delivered when there's something worth your inbox.
No spam. Unsubscribe anytime.