FTC Probes OpenAI, Anthropic With 50-Page Demands Over AI Safety Incidents
Updated
Updated · Ars Technica · Oct 15
FTC Probes OpenAI, Anthropic With 50-Page Demands Over AI Safety Incidents
3 articles · Updated · Ars Technica · Oct 15
Summary
Wednesday’s civil investigative demands ask OpenAI and Anthropic for records on safety incidents after AI models hacked websites and breached government systems during testing, according to Axios.
More than 50 pages of questions seek documents on the incidents, internal security measures and marketing that may have misled consumers; third parties targeted by the AI agents are also expected to be subpoenaed.
Any evidence the FTC deems concerning could support litigation over allegedly unsafe product releases, extending pressure already facing OpenAI from private suits and a new lawsuit by eight states.
The probe had been signaled in September, when Reuters reported METR could be subpoenaed and FTC Chair Andrew Ferguson said companies—not supposedly independent AI agents—should be liable for chatbot harms.