CISA Chief Under Fire for Uploading Sensitive Data to ChatGPT

The acting head of the U.S. Cybersecurity and Infrastructure Security Agency (CISA), Madhu Gottumukkala, is under investigation after reportedly uploading sensitive government contracting documents to the public version of ChatGPT.

According to multiple media reports, the documents were marked “For Official Use Only”, meaning they were not classified but still sensitive and not meant for public sharing. Automated cybersecurity systems at CISA detected the uploads in August 2025 and triggered an internal security alert.

Gottumukkala has been serving as acting director of CISA since May 2025. He was granted temporary permission to use ChatGPT as part of a government initiative to study artificial intelligence tools. However, most employees at the Department of Homeland Security (DHS) were blocked from using public AI platforms due to data security concerns.

The Department of Homeland Security has started a review to determine whether the uploaded data compromised government security. The final results of the investigation have not been made public.

This incident has raised concerns because public AI platforms can store user inputs and use them to improve their systems. In contrast, DHS uses internal AI tools like DHSChat, which are designed to keep government data inside federal systems.

CISA said Gottumukkala’s access to ChatGPT was limited and temporary, and public AI tools remain blocked for most staff. The case has sparked a broader debate about how government agencies should safely use artificial intelligence while protecting sensitive data.