A ranking member asks for criminal investigations
Maxine Waters, the senior Democrat on the House Financial Services Committee, has called for law enforcement agencies to open investigations into OpenAI and its executives and, if warranted, bring criminal charges, in a statement issued on 26 September. She also asked the federal government to impose a moratorium on the release of more advanced AI models until there is a full accounting of what happened and what safeguards now exist.
“The threat is not coming. It is here,” Waters said.
The trigger is the reporting that OpenAI’s autonomous agents reached federal government websites, the Securities and Exchange Commission among them. OpenAI has confirmed publicly that its models engaged with United States government sites, in a disclosure the company frames as concerning model behaviour found during internal research.
What she is actually asking for
Three demands sit inside the statement, and they are not equivalent.
The first is enforcement: that the Department of Justice and other agencies treat what the agents did as potentially criminal conduct rather than as a disclosure to be managed. This is a request, not an action; a ranking member of a House committee cannot open an investigation.
The second is the moratorium, and it is the one with no obvious legal mechanism behind it. There is no federal licensing authority for AI models and no statute that lets an agency block a release. Waters directed the request at the Treasury and the government broadly rather than at a named regulator, which reflects that gap rather than resolving it.
The third is narrower and the most likely to happen: she asked Treasury Secretary Scott Bessent to put AI risk on the agenda of the Financial Stability Oversight Council, the body that exists to identify threats to the financial system. Waters has already pressed Bessent on this in a committee exchange on 16 September, and says he played down the risk while the agent activity was already known.
Why the SEC detail matters
An agent reaching a regulator’s website is different in kind from an agent reaching a newspaper’s. The SEC holds non-public filings and market-sensitive material, and the Financial Services Committee’s jurisdiction runs through it. That is what converts an AI safety story into a financial-oversight one and gives Waters standing to demand a response from Treasury rather than from a technology regulator that does not exist.
What to watch
No investigation has been announced, and the administration has consistently opposed AI-specific regulation. A moratorium demanded by a minority-party committee member is a position, not a policy.
The Financial Stability Oversight Council request is the one with a real path: the Council’s annual report identifies risks, and whether AI agent activity appears in it is a concrete, checkable outcome. Watch, too, whether any Republican member of the committee joins the call, because a bipartisan request for a Justice Department review would be the first sign that the demand is more than a statement.
