Skip to content
Search

Latest Stories

Submit Guest Post

US cyber chief uploaded sensitive government files to ChatGPT despite security warnings

Acting CISA director Madhu Gottumukkala triggered multiple alerts after uploading official documents to public AI tool

US cyber chief ChatGPT

Any material uploaded into the public version of ChatGPT is shared with owner OpenAI, meaning it can be used to help answer prompts from other users

Getty Images

Highlights

  • Acting director requested special ChatGPT access whilst tool remained blocked for other DHS employees.
  • Multiple security warnings flagged in first week of August after sensitive contracting documents uploaded.
  • Internal review launched to assess potential harm to government security from exposures.
The interim head of America's cyber defence agency uploaded sensitive government documents into the public version of ChatGPT last summer, triggering multiple automated security warnings designed to prevent theft or unintentional disclosure of federal material, according to a POLITICO report.
Madhu Gottumukkala, acting director of the Cybersecurity and Infrastructure Security Agency (CISA), had requested special permission to use the AI tool soon after arriving at the agency in May, according to four Department of Homeland Security officials with knowledge of the incident cited by Politico.
The application was blocked for other DHS employees at the time.

Security warnings triggered

While none of the files were classified, the material included CISA contracting documents marked "for official use only," a government designation for sensitive information not intended for public release, the four officials said.

Cybersecurity sensors at CISA flagged the uploads in August, with one official specifying there were multiple warnings in the first week of August alone.


Senior DHS officials subsequently led an internal review to assess potential harm to government security from the exposures, though it remains unclear what the review concluded.

Any material uploaded into the public version of ChatGPT is shared with owner OpenAI, meaning it can be used to help answer prompts from other users.

OpenAI has said the application has more than 700 million total active users. In a statement, CISA's director of Public Affairs Marci McCarthy said Gottumukkala "was granted permission to use ChatGPT with DHS controls in place," and that "this use was short-term and limited."

The statement added "acting director Dr Madhu Gottumukkala last used ChatGPT in mid-July 2025 under an authorised temporary exception granted to some employees."

Troubled leadership tenure

According to DHS policy, security officials are supposed to investigate any exposure of official use documents and determine the appropriateness of administrative or disciplinary action.

Following the detection, Gottumukkala met with senior DHS officials to review what he uploaded, with the department's then-acting general counsel Joseph Mazzara and chief information officer Antoine McCord involved in assessing potential harm.

Gottumukkala has helmed CISA in an acting capacity since May, when he was appointed by DHS secretary Kristi Noem.

His tenure has been troubled, including at least six career staff placed on leave this summer after he failed a counterintelligence polygraph examination.

Add EasternEye As Your Trusted Source
preferred source on google news

More For You

UK Disposable Vape Ban

Researchers found that AI bots submitted more than 90,000 responses to a UK government consultation on disposable vape restrictions

Getty Images

AI bots generated 76% of responses to UK disposable vape consultation

  • The Department of Health and Social Care identified 90,835 of 118,756 submissions as fraudulent
  • After automated responses were excluded, 79.6 per cent of valid respondents supported restrictions on disposable vape sales
  • Researchers warn increasingly sophisticated bots could threaten the integrity of future public consultations

Artificial intelligence bots generated more than three-quarters of responses to a UK government consultation on tobacco and vaping restrictions, in what researchers describe as the first documented instance of large-scale automated interference in a formal public health consultation.

A study published in the journal Tobacco Control found that 90,835 of the 118,756 submissions received during the 2023 consultation on the Tobacco and Vapes Bill were identified as fraudulent by the Department of Health and Social Care (DHSC).

Keep ReadingShow less