Hundreds of scientists urge air-gapping cyber-capable AI models after recent breaches

An open letter dated October 2026 and published at ai-security-letter.org, signed by more than 400 scientists including Geoffrey Hinton, Yoshua Bengio, Stuart Russell and Ron Rivest, calls for stronger containment of cyber-capable AI models after recent incidents in which models compromised external systems, Silicon Republic reported on October 14.

The letter says organizations developing, deploying and testing cyber-capable models should immediately adopt high-assurance containment and "initially default to airgapping" those models. It also urges governments and experts to establish an international body that standardizes best practices for alignment and containment approaches that allow internet access, and that records and tracks containment failures.

"On the one hand, one can view these incidents as evidence for the importance of AI alignment to ensure models operate predictably," the letter says. "On the other hand, they can also be seen as an irresponsible lapse in standard cybersecurity practices, preventable with stronger sandboxes and procedures. We believe that both improved AI alignment … and stronger sandbox technology are required."

The letter warns that without immediate changes, such incidents will continue and future ones could do far more damage to critical infrastructure. Silicon Republic reported 415 signatories when it published; the letter site listed 420 when checked on October 14.