Generative AI Terrorism Risk Assessment Act

The bill's focus shifted from 'terrorist organizations utilizing generative AI applications' to 'foreign terrorist organizations' and individuals using generative AI for terrorism. The definition of 'terrorist organization' was narrowed to only include entities designated as FTOs under the Immigration and Nationality Act, removing the broader category of any entity engaged in terrorism.

The deadline for the first annual assessment changed from 180 days after enactment to one year after enactment, giving DHS more time to prepare.

A new coordination requirement was added to the information-sharing section specifying that the Office of the Director of National Intelligence, the FBI, members of the intelligence community, and any other relevant federal agency determined necessary by the Secretary must share relevant information with DHS.

New definitions were added for 'artificial intelligence' (referencing the National AI Initiative Act of 2020), 'terrorism' (referencing the Homeland Security Act), and 'intelligence community' (referencing the National Security Act). The term 'generative artificial intelligence applications' was simplified to 'generative artificial intelligence.'

The publication rules for assessments were tightened: the unclassified portion must not contain information that is FOIA-exempt or otherwise not publicly available. The restriction on classified annexes (previously limited to protecting intelligence sources and methods) was removed, allowing a broader basis for classification.

The coordination requirement in the assessment process was simplified from requiring coordination with specific DHS offices (General Counsel, Privacy Office, Civil Rights and Civil Liberties) to a general requirement that the Secretary ensure compliance with applicable law and privacy/civil rights protections. References to 'State and local fusion centers' were changed to 'State and major urban area fusion centers.'