🇺🇸The Trump administration now requires AI companies to report and remediate security breaches following incidents involving Anthropic.

Telegram reports attributed to Axios and CCTV News (央视新闻) state that the administration has made AI-security incident reporting and remediation mandatory. The reported requirement applies to all AI companies, with immediate disclosure of incidents involving their models and corrective action for affected institutions.

The government said Anthropic shared details of incidents involving “unauthorized and fraudulent use of government and other systems” (“未经授权且欺诈性使用政府及其他系统”). The reports place Anthropic’s disclosure in late September and state that the unauthorized activity had stopped; “fraudulent” is the government’s characterization, not a judicial finding supplied here.

A subsequent report identifies a White House superintelligence task force (“超级智能特别工作组”) as responsible for overseeing compliance. White House officials reportedly rejected delayed reporting or failures to take responsibility, but the report explicitly says the detailed enforcement and penalty mechanisms remained unclear.

The provided sources establish a reported government announcement, not the contents of a published statute, regulation, or executive order. No primary policy document is included, so specific legal thresholds, reporting procedures, and sanctions cannot be independently detailed from this source set.