In mid-July 2025, CISA’s acting director Madhu Gottumukkala uploaded government contracting documents marked “for official use only” into the public, consumer version of ChatGPT, triggering automated security alerts inside the Department of Homeland Security. He had obtained special permission to use commercial AI tools shortly after taking the role, even though ChatGPT was otherwise blocked for most DHS employees over data-retention concerns. The episode, first reported by Politico and confirmed by multiple outlets in late January 2026, prompted a DHS security review and a congressional inquiry into CISA’s AI governance.