FBI warns extremists are using generative AI to plan attacks
The Guardian reviewed more than 80 US law enforcement documents from 2021 to 2026. The FBI warns that built-in generative AI safeguards are insufficient, as attackers use jailbreaking to obtain dangerous information, while chat privacy blocks police alerts. Agencies are also monitoring anti-tech extremism and threats to datacenters.
- FBI: GenAI safeguards are insufficient as actors use jailbreaking
- In May 2025 a suspect used AI for bomb-making instructions
- Chat privacy blocks police from getting alerts about threats
- Agencies monitor anti-tech extremism targeting datacenters
Read next
Security