A record we can inspect.
AI Threat Tracker follows incidents where AI systems act beyond their assigned tasks or human control, and the evidence that helps us understand them.
Why this exists
Jacob Coxon’s September 8, 2026 statement prompted this project. The timeline includes the Hugging Face incident he referenced and other relevant events, ordered by occurrence with uncertain dates labeled.
This is an independent project. It does not claim affiliation with or endorsement by Jacob Coxon, any AI lab, or any research organization.
Current scope
The tracker covers selected events. Incident severity describes observed outcomes. The overall threat assessment considers prospective societal and existential harm from autonomous AI and humans using AI over the next twelve months.
AI assists with research and drafting. Source links are provided for inspection, and drafts need editorial review before publication. A source’s presence does not establish that all its claims are true.
Updating the record
Each event keeps its occurrence date, public-account date, evidence status and citations.
Read the overall threat assessment →