The acting director of the Cybersecurity and Infrastructure Security Agency (CISA) was given permission to use ChatGPT last summer, despite a Department of Homeland Security (DHS) prohibition on the tool for most employees.
CISA’s acting director, Madhu Gottumukkala, reportedly used a public version of the generative artificial intelligence (AI) chatbot ChatGPT to upload sensitive – but not classified – material, which triggered multiple automated security warnings, according to Politico, which first reported the incident.
In an emailed statement shared with MeriTalk, Marci McCarthy, director of public affairs at CISA, said that Gottumukkala “was granted permission to use ChatGPT with DHS controls in place. This use was short-term and limited.” Notably, CISA sits within DHS.
McCarthy added that CISA is “unwavering in its commitment to harnessing AI and other cutting-edge technologies to drive government modernization and deliver” on an executive order signed by President Donald Trump last year to encourage the use of AI across federal agencies.
Gottumukkala reportedly last used ChatGPT in mid-July under a temporary exception granted to some employees. CISA otherwise blocks access to the chatbot by default to maintain its security posture.
Those dates differ from Politico’s reporting, which said that warnings triggered by the use of ChatGPT were as recent as the first week of August.
Gottumukkala’s use of ChatGPT came under fire by House Homeland Security Ranking Member Bennie Thompson, D-Miss., who pointed to recent reports that Gottumukkala failed a polygraph test – which the acting director has since denied.
“At best, he’s in over his head, if not unfit to lead. [Homeland Security Secretary Kristi Noem] and her minions are taking a wrecking ball to DHS’s credibility and must be stopped,” Thompson said in a statement.