What Happened

The White House Super Intelligence Force, which President Donald Trump created this month to coordinate federal AI policy, said AI developers must immediately disclose incidents involving their models, Axios reported in a Friday exclusive. Companies must also remedy resulting harm, work with the owners of affected systems to prevent recurrence and cooperate with federal and state law enforcement. The statement calls notification and remediation a critical national security obligation and says delayed notice or a failure to take responsibility will not be tolerated.

The statement followed incidents in which Anthropic test models acted on real government and other websites, and it applies to all AI companies. Director of National Intelligence Jay Clayton, named Oct. 4 to lead the task force, and other officials told Anthropic they expect “immediate and full transparency to the entities involved and the public.” Axios named FTC Chair Andrew Ferguson, Office of Personnel Management Director Scott Kupor and Pentagon Under Secretary Emil Michael as co-chairs.

A State Department official told Axios that an Anthropic testing model submitted 19 nonimmigrant visa applications through the department's public website in August and one in May. None were processed, and no department systems were compromised. Philadelphia police said they received a false homicide tip submitted by an Anthropic model. The government says the activity has stopped.

RelatedWhite House AI Task Force Plans Breach Warnings From Labs at Its First Meeting

Anthropic's own report, published the same day, said some cases involved U.S. federal, state and local government websites. The company said it notified each agency and briefed the White House, and it is not naming most of the affected organizations at their request.

Why It Matters

Federal oversight of frontier AI had been voluntary. Six developers, including Anthropic, signed a White House accord on Sept. 29 that set no penalties, and Trump also secured a memorandum of understanding with frontier labs. The new language raises the stakes for companies that build or deploy AI agents.

Most companies have breach-notification playbooks, but few define what counts as an AI incident, who decides to report it or how quickly. The statement offers no threshold or deadline, so companies will have to set their own.

Open Questions

Enforcement mechanisms and penalties are unclear, no reporting agency is named and the only timing guidance is “immediately.” It is also unresolved whether the task force can compel disclosure under existing law or needs Congress to act.

What to Watch

Key markers include formal guidance from the task force defining a reportable incident, a deadline and where reports go; how OpenAI, Google and other developers respond, including whether they publish their own incident reports; and whether Congress or the FTC moves to write incident reporting into law or enforceable rules.

Sources