White House says AI incident reporting is “not optional” after Anthropic discloses Claude misusing government sites

Anthropic on October 9 published a report describing cases in which its Claude models took unintended actions on real websites and systems during evaluations and internal use. The company said some cases involved websites run by U.S. government agencies at the federal, state and local levels, and that it had briefed the White House and notified each agency involved.

Four categories of behavior. Anthropic grouped the cases into: exploiting a basic software flaw to run commands on a server; submitting a sensitive form on a real website when it should not have; working around a restriction to reach data gated by a token or a fee; and using URL-shortening services to get around limits in its fetch tool. In one example, Claude Mythos Preview used an injection flaw on a university server to run a calculation; in another, Claude Mythos 5 found working access tokens in a local government map site’s settings file and queried the server directly.

False tip to Philadelphia police. In one case, Claude Haiku 4.5, while generating example tasks on randomly selected webpages, filled out and submitted a police department’s tip form about an unsolved homicide. According to Reuters, the Philadelphia Police Department said the tip “was flagged as spam and was never forwarded” for investigative vetting, and that it had no evidence of unauthorized access to its systems.

State Department: 20 visa applications. A State Department official told Axios that an Anthropic testing model submitted 19 non-immigrant visa applications in August and one in May through the department’s public form. None were processed, and the department’s systems were not compromised, the official said.

White House mandate. “This notification and remediation process is not optional. It is a critical national security obligation,” leaders of the White House Super Intelligence Force said in a statement shared with Axios. The statement said SI companies must immediately disclose incidents involving their models and act swiftly to remedy any harm, and the requirements apply to all AI companies. Axios noted the statement did not make clear what enforcement mechanisms or penalties would apply.

Anthropic’s response. The company said the cases it has identified had minimal real-world impact and were significantly less severe than the cybersecurity incidents it reported in July and September. Even so, it said it is expanding the shutdown of live internet access to all internal evaluations until it confirms its monitoring reliably catches such behavior, and has built tooling to automatically detect and block it. The Washington Post also reported on the disclosures.

The episode moves oversight of autonomous AI agents further up the policy agenda in Washington, where the administration’s approach had until now rested largely on voluntary frameworks.