OpenAI faced legal and regulatory pressure on both sides of the Atlantic in a single week, with eight US states suing the company over allegations that it misled the public about product safety. Separately, the Federal Trade Commission reportedly subpoenaed OpenAI and Anthropic over consumer harms, while European officials announced information requests concerning AI security and oversight.
The state legal action came on Wednesday. Its central allegation concerns how OpenAI represented the safety of its products to the public. Those assertions remain allegations, not findings of wrongdoing. The available source material does not identify the eight states, detail the remedies they are seeking or include OpenAI’s response.
On Thursday, Axios reported that the FTC had subpoenaed both OpenAI and Anthropic in connection with consumer harms. The reported federal inquiry is distinct from the states’ lawsuit. The available account does not specify which products or practices the subpoenas concern, what records the agency requested or whether the inquiry will lead to enforcement action.
A subpoena is an investigative tool used to compel information; it does not itself establish that a company has violated the law. Consumer-protection investigations can examine whether businesses’ public representations match their practices, as well as whether products expose users to unlawful or inadequately disclosed risks. Any eventual enforcement case would depend on the evidence gathered and the applicable legal standards.
In Europe, European Commission executive vice president Henna Virkkunen announced on Wednesday that the AI Office had sent requests for information covering security, external evaluation and monitoring of how models are used. The source material does not provide the full recipient list or the text of those requests, limiting what can be concluded about their specific scope.
The subjects identified by the Commission involve different stages of AI oversight. Security concerns how systems are protected against threats and misuse. External evaluation provides scrutiny beyond a developer’s own testing, while monitoring can help identify problems that emerge once models are deployed. Testing before release and observing use afterward address different questions: a model’s behavior under controlled conditions may not capture every risk encountered in real-world applications.
The three developments also involve different legal mechanisms. A lawsuit asks a court to resolve allegations and potentially order relief. An investigative subpoena seeks evidence. A regulatory information request gathers material needed for supervision or further assessment. None should be treated as interchangeable with a final determination of liability.
For now, the public account leaves important procedural details unresolved, including the states’ requested relief, the scope of the reported FTC demands and the recipients of the European requests. The developments nevertheless place OpenAI’s safety representations and oversight practices under scrutiny through several separate channels, without establishing the outcome of any of them.
Sources: Tech Policy Press, Tech Policy Press, OpenTools