Timeline of Warnings of AI Risks
Public warning about AI riskLab safety actionIncident (colored by type, see the incident timeline below)
Geoffrey Hinton’s Nobel Prize banquet speech, Stockholm, Dec. 10, 2024 (starts at 2:08).
Yoshua Bengio, “Speech at the UN Security Council on Frontier AI Risks & Global Governance,” Sept. 23, 2026.
Timeline of Reported Incidents
Sandbox escape
Misconfigured eval environment
Authorized Internet, unsanctioned actions
Escape during training
Self-modification (test only)
Live deployment, unsanctioned action
First public disclosureLater report
Most serious (our assessment)
Time until first disclosure