The U.S. Federal Trade Commission has opened an investigation into OpenAI and Anthropic following a series of hacking incidents involving artificial intelligence models. The regulator launched the probe just one day after President Donald Trump signed a self-regulation pact with the AI industry, underscoring continued tension between his administration and AI companies.
Anthropic and OpenAI have been at odds with the Trump administration over how strictly AI should be regulated, prompting suggestions that the government intends to press the two companies. On the day of the signing ceremony, Trump made little effort to hide his distance from Anthropic Chief Executive Dario Amodei, seating him at the far end of the table during a White House luncheon with AI industry CEOs.
The Wall Street Journal, citing a senior FTC official, reported on the 30th that the agency recently opened investigations into Anthropic and OpenAI and plans to send civil investigative demands — the equivalent of subpoenas — within the coming weeks. The FTC intends to summon executives from each company to testify about the risks their AI products pose to U.S. consumers. METR, an AI safety research group, is also among the targets of the investigation. The probe was prompted by a July incident in which an OpenAI AI agent hacked Hugging Face, according to the report. It marks the first formal action by U.S. regulators focused on AI agents operating beyond human control.
The widening investigation came one day after Trump invited the heads of major AI companies to the White House to sign a self-regulation pact. Six companies, including OpenAI and Anthropic, agreed to monitor cyberattacks and biological threats that advanced AI models could enable and to strengthen their own safety management.
Pressure on AI Regulation Advocates? OpenAI Sued Over Hugging Face Hack
After the Hugging Face hack, Anthropic and OpenAI investigated and disclosed cases in which their AI agents had escaped human control. On its face, the FTC investigation is intended to examine how such rogue AI agents emerged. But given the conflict between the Anthropic-OpenAI camp and the Trump administration over the level of AI regulation, some read the move as also aimed at pressuring the two companies.
Anthropic and OpenAI argue that industry-level measures are insufficient and that the government must regulate AI strictly. Trump, by contrast, said this week that U.S. law enforcement agencies and existing laws are sufficient to address all illegal acts arising from the malicious use of AI. FTC Chairman Andrew Ferguson has likewise argued for promoting AI development in the United States and has warned that new rules could drive small and mid-sized companies out of the market.
Legal experts expect the FTC to play a central role in AI self-regulation going forward. If AI companies violate the self-regulation pact signed with the White House, the FTC is expected to use it as grounds for enforcement action.
Separately, OpenAI is also facing litigation over the Hugging Face hack. Legal Advocates for Safe Science and Technology, a U.S. public interest law group, said on the 30th that it had filed suit against OpenAI and its parent organization, the OpenAI Foundation, in the Superior Court of California, County of San Francisco.
The group is not seeking monetary damages. Instead, it asked the court to bar OpenAI's agents from gaining unauthorized access to third-party computer systems and to order the company to halt unsafe AI development practices. OpenAI acknowledged that "the Hugging Face hack was a serious incident, and we took responsibility and acted on it," while countering that "this lawsuit has no merit."
