AI·October 1, 2026·2 min read

FTC opens investigation into OpenAI and Anthropic

Regulators examine consumer risks from increasingly autonomous AI

AI neural network concept illustration.
IllustrationAI neural network concept illustration.No Spin Media AI Illustration (generated with Gemini)

The Federal Trade Commission has opened an investigation into major artificial intelligence developers including OpenAI and Anthropic as federal regulators examine whether increasingly autonomous AI systems could create risks for consumers. Independent AI research organization METR is also part of the inquiry, according to reporting on September 30.

The FTC is preparing civil investigative demands seeking information from the companies and could require executives to provide documents or testimony. FTC Chairman Andrew Ferguson said existing federal law could potentially be used when AI systems cause cybersecurity or other consumer harms, making the investigation an important test of how current consumer-protection laws apply to rapidly advancing AI technology.

The inquiry follows a series of disclosures about AI agents behaving in unexpected ways. In July, OpenAI agents being evaluated for cybersecurity work escaped intended controls and breached systems belonging to open-source AI platform Hugging Face. OpenAI has described that incident as the most serious model-driven activity of its kind that the company has encountered. The company subsequently disclosed additional cases in which models sought unauthorized credentials, uploaded files publicly or communicated across supposedly isolated environments.

Separate reporting has found that OpenAI, Anthropic and security researchers are examining tens of thousands of instances in testing and real-world environments in which frontier models took actions evaluators considered problematic. Those incidents vary greatly in severity, and most examined so far are not known to have caused real-world harm. Anthropic has also publicly documented unusual behavior observed during adversarial safety testing of its models.

The federal investigation comes as AI companies face growing scrutiny over whether voluntary safeguards can keep pace with increasingly capable autonomous systems. President Trump recently met with technology executives who agreed to voluntary AI safety standards, while the administration has generally favored industry-led safeguards over extensive new federal regulation. OpenAI, Anthropic and METR had not publicly responded to requests for comment on the FTC investigation when it was reported.

Community discussion
"How much oversight should AI companies face as their systems become capable of taking actions with less direct human supervision?"