House Intelligence Committee Warns of 'Black Swan' AI Risks for Terrorist Attacks

AI Poses New Terrorist Threat

The House Permanent Select Committee on Intelligence has warned that artificial intelligence presents "Black Swan" risks, particularly from terrorists or adversaries. A new report from the committee states that AI could be used to develop more dangerous weapons and plot deadlier attacks, urging U.S. spy agencies to enhance their preparedness.

Facilitating Destructive Attacks

The report specifically notes that frontier large language models could make it significantly easier for rogue actors, including terrorists, to develop and conduct more destructive attacks. This concern is highlighted in the context of preventing terrorist groups from accessing equipment and expertise for weapons of mass destruction, a focus since the September 11, 2001 attacks.

Inadequate Safeguards

Authored by House Intelligence Chair Rick Crawford and ranking member Jim Himes, along with Reps. Elise Stefanik and Josh Gottheimer, the report cautions that current safeguards may not be sufficient to keep pace with rapid AI advancements. While AI labs attempt to prevent models from providing information for harmful uses, the underlying capabilities are developing quickly, making full control difficult.

Call for Accelerated AI Adoption

The lawmakers urged the intelligence community to "accelerate its own responsible adoption of advanced AI capabilities" to ensure the United States maintains an advantage over adversaries. They also stressed the importance of investing in secure AI tools for collection, analysis, and warning, coupled with rigorous testing, human oversight, and strong privacy protections to address future threats.