WASHINGTON (Web Desk) - The acting head of the United States’ cyber defence agency uploaded sensitive government documents to the public version of the AI chatbot ChatGPT, raising fresh concerns about data security within federal agencies, according to a Politico investigation.
The report revealed that Madhu Gottumukkala, Acting Director of the Cybersecurity and Infrastructure Security Agency (CISA), received special permission from CISA’s Office of the Chief Information Officer to use ChatGPT, despite the tool being blocked for all Department of Homeland Security (DHS) officials at the time.
In 2025, Gottumukkala reportedly uploaded agency documents to the chatbot, triggering internal alarms over a potential data breach.
Four DHS officials familiar with the matter said the uploads activated multiple automated security alerts designed to prevent the unauthorised transfer or accidental disclosure of government information.
While the documents were not classified, they were labelled “For Official Use Only,” a designation that still requires strict handling and protection, according to officials.
Following the incident, the Department of Homeland Security initiated an internal review to assess the scope of the exposure and any potential damage to national security.
However, it remains unclear what conclusions the review reached or whether any disciplinary action was taken.
In response to the incident, CISA has since developed its own internal AI-powered chatbot, known as DHSchat, aimed at allowing the use of generative AI tools while preventing sensitive information from leaving federal networks.
Who is Madhu Gottumukkala?
Madhu Gottumukkala is an Indian-origin American official currently serving as Acting Director and Deputy Director of the Cybersecurity and Infrastructure Security Agency.
He has more than 24 years of experience in information technology and holds academic qualifications in engineering, computer science, business management and information systems.