CISA's Acting Director Faces Backlash Over Sensitive Data Upload

Madhu Gottumukkala, the acting leader of the U.S. Cybersecurity and Infrastructure Security Agency (CISA), finds himself under investigation after allegedly uploading sensitive government documents to the public version of ChatGPT.

Reports indicate that these documents, labeled “For Official Use Only”, were not technically classified but were still sensitive and intended for limited distribution. In August 2025, automated cybersecurity systems at CISA identified the uploads, leading to an internal security notification.

Gottumukkala has filled the position of acting director of CISA since May 2025, allowed to use ChatGPT as part of a governmental experiment with artificial intelligence. However, most personnel at the Department of Homeland Security (DHS) have been restricted from accessing public AI platforms due to serious data security risks.

DHS has initiated an inquiry to assess whether the uploaded data jeopardized government security, though the investigation's findings have yet to be released.

This occurrence raises red flags, as public AI platforms can retain user data and utilize it for system enhancement. In contrast, DHS employs internal AI solutions like DHSChat, specifically designed to secure government data within federal systems.

CISA has clarified that Gottumukkala's access to ChatGPT was both restricted and time-limited, with most staff continuing to be banned from using public AI tools. This incident has ignited a wider discussion on how government entities can responsibly leverage artificial intelligence while safeguarding sensitive materials.