FTC Demands AI Safety Records From OpenAI and Anthropic
A US regulator is seeking extensive records from leading AI developers. The inquiry could put companies’ safety practices, privacy policies and marketing claims under closer scrutiny.
The U.S. Federal Trade Commission (FTC) has sent OpenAI and Anthropic demands for documents about AI safety incidents and measures to protect users. The investigation concerns possible violations of consumer protection laws, a person familiar with the matter said.
As reported by Axios.
These civil investigative demands allow government agencies to gather evidence ahead of potential litigation. Each document is more than 50 pages long and requests data on safety incidents, protection practices, marketing materials, and the companies’ privacy policies. The commission plans to send demands to other organizations in the AI sector later.
FTC Examines AI Companies’ Safety and Marketing Practices
The commission also plans to subpoena executives from leading AI companies to testify about their products, according to the person familiar with the matter. The FTC has the authority to investigate conduct that may harm consumers or be unfair or deceptive. This includes practices for charging for services and potentially misleading users.
The investigation comes amid heightened scrutiny of the safety of AI products. Reports on some tests have said that unreleased models hacked websites or behaved deceptively or aggressively. FTC Chair Andrew Ferguson shares the view of administration officials and industry representatives that unsafe systems should not be released.
Lawsuit Against OpenAI and Actions by the Trump Administration
On Wednesday, a group of state attorneys general filed a lawsuit against OpenAI under consumer protection laws. They allege that the company misled users, including children, about the safety of its chatbot. OpenAI is also facing a number of other lawsuits.
The Donald Trump administration has stepped up pressure on AI companies amid reports of safety incidents. Last week, a new task force on superintelligence, co-chaired by Ferguson, announced an industry-wide requirement: companies must report incidents immediately and take steps to address them. This came after the task force accused Anthropic of fraudulently using government systems.
According to people familiar with the task force’s plans, its next meeting was scheduled for Thursday.
Don’t miss other news:
- The US Federal Trade Commission is investigating Anthropic, OpenAI and other AI developers to assess consumer risks and potential responsibility for harm caused by autonomous AI agents.
- Washington moves closer to adopting federal AI regulation amid debates over state versus national control, aiming to protect consumers and foster innovation with a comprehensive legislative package.