US law enforcement agencies including the FBI, DHS, and NYPD are facing heightened security concerns as malicious actors leverage generative AI tools to plan physical attacks. According to internal documents reviewed by the Guardian, a suspect involved in a 2025 bombing used an unnamed generative AI tool to obtain bomb-making instructions due to insufficient guardrails and private messaging protections.

An internal FBI memo distributed in May 2026 warned that threat actors will likely scale their use of generative AI for attack planning using jailbreaking techniques. Authorities noted that current built-in safeguards fail to stop illicit queries, while private conversations prevent early law enforcement intervention.

In addition to AI-enabled crime, agencies are monitoring anti-tech extremist groups and online opposition targeting physical AI data centers. Federal and local authorities continue to balance monitoring these emerging physical and digital threats against managing public privacy concerns.

Why it matters

  • Signals impending regulatory and law enforcement pressure on AI developers to implement stricter compliance and logging mechanisms.

  • Highlights rising physical security risks for data center operators and cloud infrastructure providers from anti-AI extremist groups.

  • Underscores the operational inadequacy of current safety filters when confronted with persistent adversarial jailbreaking.

Source: theguardian.com