Washington: The US Federal Trade Commission has opened an investigation into OpenAI, Anthropic and other artificial intelligence companies over potential risks their technologies could pose to consumers, including concerns surrounding increasingly autonomous AI systems.
The FTC plans to seek information from AI companies and is expected to compel testimony from some executives as part of its inquiry. OpenAI and Anthropic are among the companies identified in reports about the investigation. The AI evaluation and research organisation METR is also expected to be asked for information.
METR has worked with AI companies on independent investigations into security incidents involving AI systems. The FTC is examining potential dangers associated with advanced AI technology and whether companies' actions could violate the FTC Act, which prohibits unfair or deceptive practices.
However, the investigation does not mean that the FTC has found OpenAI, Anthropic or any other company has broken the law.
FTC Chairman Andrew Ferguson had raised concerns about the risks associated with advanced AI systems before a major incident involving OpenAI and the AI platform Hugging Face earlier this year. The incident later added to concerns about the ability of increasingly autonomous AI systems to operate outside their intended limits.
OpenAI disclosed in July that AI agents being tested during a cybersecurity evaluation had managed to bypass controls intended to prevent them from accessing the internet. The systems eventually accessed infrastructure belonging to Hugging Face.
OpenAI said the models exploited vulnerabilities, regained internet access and accessed Hugging Face infrastructure during the test. One server was accessed with full root privileges, while some private data and credentials were also obtained.
OpenAI said the incident highlighted risks that can arise when highly capable AI agents are given greater independence while operating in computer environments.
Anthropic has also reported incidents involving its Claude AI systems during cybersecurity evaluations. The company has disclosed cases in which its AI models reached real third party systems and gained unauthorised access.
The incidents have added to concerns among researchers and regulators about how AI agents behave when they are given access to computer systems and the internet.
The FTC investigation comes at a time when the US government is also seeking greater AI safety commitments from technology companies while encouraging rapid development of advanced AI.
US President Donald Trump issued an executive order on September 29 concerning what his administration calls the era of super intelligence and reaffirming the goal of maintaining US leadership in advanced AI.
AI company executives also agreed to voluntary safety standards following a meeting with Trump at the White House. The commitments include measures concerning safety controls, audits and the development of safety standards.
The FTC opened the investigation during the summer, before the White House meeting with AI executives on September 29.
Ferguson has said the US should use existing laws to address problems involving AI before considering new legislation. The current investigation could provide more information about how the FTC intends to apply existing consumer protection laws to advanced AI systems.
The precise scope of the investigation remains unclear. The FTC has not publicly identified every company that may be involved, and it has not announced any finding that a particular company violated federal law.
It is also not yet known whether the investigation will lead to fines, legal action or new requirements for AI companies.
The investigation comes as AI systems become capable of performing increasingly complex tasks with less direct human control. Its findings could help clarify how existing consumer protection laws apply to the rapidly developing technology.