Madhu Gottumukkala, the acting director of the U.S. Cybersecurity and Infrastructure Security Agency (CISA) with Indian roots, is under fire after reportedly uploading sensitive government documents to the public version of ChatGPT. The incident came to light last summer when Gottumukkala, who started at CISA in May 2025 under President Trump's administration, sought and received special access to the AI chatbot despite a department-wide block on it. Agency systems flagged multiple uploads of "for official use only" files—contracting docs tied to CISA operations—in early August 2025, prompting an internal Department of Homeland Security (DHS) probe.

Sources describe Gottumukkala as pushing hard for "short-term and limited" ChatGPT approval, complete with safeguards from CISA's tech team. One DHS insider quipped to reporters that he essentially "forced access, then abused it," highlighting the irony of a top cyber official bypassing standard protocols. While the files weren't top-secret, their sensitive label means they weren't for public consumption. Feeding them into ChatGPT's free tier raises alarms since OpenAI could incorporate the data into training or even surface it in responses to its vast user base.

CISA spokesperson Marci McCarthy confirmed the temporary permission but noted the review is standard for any potential data mishaps, which could lead to disciplinary action. No final outcomes have been shared, and Gottumukkala's last known use was mid-July 2025. This episode spotlights escalating tensions around AI in government. Federal agencies grapple with tools that supercharge work but threaten leaks—echoing cases like Samsung's 2023 ban on ChatGPT after code spills or chatbots giving bogus advice. With Trump's pro-AI directives aiming to slash red tape, CISA's role in securing infrastructure just got more scrutinized.