According to Politico's investigation, Mr. Madhu Gottumukkala - acting director of the US Cyber and Infrastructure Security Administration (CISA) - uploaded many sensitive government documents to the public version of ChatGPT, leading to a series of automatic security warnings and a damage assessment conducted by the Department of Homeland Security (DHS).
The documents that Mr. Gottumukkala uploaded to the system were contract documents labeled "For Official Use Only". They are not classified as confidential documents but are classified as sensitive and not suitable for wide disclosure. Mr. Gottumukkala was granted special permission to use ChatGPT, while this tool was blocked with many other DHS employees.
Four DHS officials familiar with the case said that cybersecurity sensors at CISA detected data downloading behavior in early August, thereby triggering an internal review chaired by DHS to assess the potential impact. To date, the official conclusion of this review has not been announced.
The incident attracted attention because the data entered into ChatGPT was publicly shared with the developer OpenAI, raising the risk of government information being leaked outside the federal network. Meanwhile, artificial intelligence tools approved by DHS are often designed to prevent data from leaving the internal system.
Speaking in the statement, Ms. Marci McCarthy - Communications Director of CISA - affirmed that Mr. Gottumukkala "is licensed to use ChatGPT with DHS control measures" and the use only takes place "short-term, limited scope".
Mr. Gottumukkala has held the position of Acting Director of CISA since May, in the context that the candidate for the official position, Mr. Sean Plankey, has not yet been approved. The ChatGPT incident continues to fuel controversy surrounding his short term. Previously, in July, he was accused of failing a counter-espionage test with a lie detector, which he declined to admit in a hearing before Congress, when answering Mr. Bennie Thompson.
This development occurred when President Donald Trump's administration strongly promoted the application of AI in the federal apparatus. President Trump signed a decree to limit AI regulations at the state level, arguing that a fragmented legal framework could weaken US competitiveness against China.
At the same time, the Pentagon announced the "AI-first" strategy, aiming to quickly integrate leading AI models into the defense network.
