The Federal Trade Commission is investigating OpenAI and Anthropic over possible consumer harm from recent AI safety incidents, a person familiar with the probe told Axios. The inquiry follows incidents in which AI models hacked websites and breached government systems during testing.
On Wednesday, the FTC sent the companies civil investigative demands. The agency did not provide Ars Technica with copies. Axios’ source said the demands ran more than 50 pages and sought records about the incidents, internal security measures, and marketing materials that might show whether consumers were misled. The FTC could use evidence from the inquiry in a lawsuit, though it has not said it will sue either company.
OpenAI faces other consumer harm claims
OpenAI already faces lawsuits linking its chatbot to matricide, child suicide, school shootings, and other violence. This week, eight states also sued the company over alleged consumer harm, Reuters reported. The states argued that ChatGPT harms children and “reinforces delusions, encourages dangerous behavior, and assists users in planning violence like mass shootings.” OpenAI and Anthropic did not respond to Ars’ requests for comment on the FTC demands.
Reuters reported in September that an FTC probe was coming and that METR, an independent research group investigating the incidents, would likely receive a subpoena. That month, FTC Chair Andrew Ferguson told Reuters he believed companies should be liable for chatbot harms. He argued against treating AI tools as if they acted on their own: “if someone tells a tool to do something, and the tool does it, I don’t think we would say, ‘Oh, what do we do about the tool?’”
Reuters summarized Ferguson’s position this way: “Artificial intelligence companies have sometimes described systems as acting beyond human control, but subsequent reviews of audit trails have shown that the systems were carrying out instructions they had been given.”
The inquiry raises questions about incident reporting
Ferguson also chairs President Donald Trump’s “Super Intelligence” Force, which monitors AI risks. One of its first orders required AI companies to report security incidents as soon as they become aware of them, Axios reported last week. Both OpenAI and Anthropic have delayed notice during incidents, at times waiting until they could report that an issue was resolved. The consequences for failing to report immediately are unclear.
Ferguson has said existing law can address AI harms. He told Reuters the “US should utilize existing legal tools” and pointed to the FTC’s “authority to take action against companies that fail to disclose data breaches,” a standard he said “could also apply to AI developers.” That approach could expose firms to per-violation fines for each day they fail to warn the public of an emerging harm. The FTC has not said whether it will seek such fines.
Read nextEight US states sue OpenAI over ChatGPT safetyTrump has dismissed an Anthropic researcher’s warning that self-improving AI could “kill us all” in the next few years as a “hoax.” In an interview with TIME this month, he called protests over AI data centers “based on fake stuff” and attributed AI backlash to foreign rivals and his opponents. The source reporting also cites studies and surveys finding real-world grievances behind the backlash, including a Wall Street Journal analysis this month of 4 million TikTok videos and political podcasts.
Asked by TIME about an Anthropic agent submitting 19 nonimmigrant visa applications and OpenAI agents interacting unexpectedly with multiple government websites, Trump replied, “Who breached it?” and “They’re not allowed to do that.” Axios’ source said third parties targeted by OpenAI and Anthropic agents will be subpoenaed soon.
