Scientists call for international body to oversee advanced cyber-capable AI systems
Around 400 scientists, including AI pioneers Geoffrey Hinton and Yoshua Bengio, published an open letter calling for an international body with authority over containment standards and security incident reporting for frontier AI models.
On October 13, 2026, a group of roughly 400 scientists from 34 countries published an open letter calling for the creation of an international body to oversee advanced cyber-capable AI systems. The signatories include AI pioneers Geoffrey Hinton and Yoshua Bengio, both of whom have been increasingly vocal about risks from frontier AI models.
The letter’s demands are specific and far-reaching. The proposed body would set industrywide containment standards, require frontier AI models to be air-gapped, and maintain independent oversight of systems with significant cyber capabilities. It would also create a registry of security incidents and publish records of containment failures — a level of transparency that does not currently exist for the most powerful AI systems in development.
The exact number of signatories and countries involved has been reported differently across sources. Quartz and The Spokesman-Review both cite around 400 scientists from 34 countries, while ABC News reported 350 scientists from 31 countries — a discrepancy that may reflect either different counting methodologies or additional signatures added after initial publication. Either figure represents a substantial coalition of researchers across the field.
The core technical proposal — air-gapping cyber-capable models — reflects a growing consensus in AI safety circles that the most capable systems should be physically isolated from networks as a baseline precaution. The letter also calls for mandatory reporting of security failures, which would create a public record of incidents currently handled privately by individual labs.
What distinguishes this intervention is its institutional ambition. Rather than calling for voluntary commitments or national regulation alone, the scientists are proposing a body with genuine international authority — something that does not currently exist for AI governance. The letter does not specify how such a body would be constituted or funded, leaving those questions to the diplomatic process it is attempting to catalyze.
The timing is notable. Hinton and Bengio have both shifted from technical research to advocacy in recent years, and their joint signature on this letter signals that concern about catastrophic risks has moved from the margins of the field toward its most decorated practitioners. Whether that translates into political momentum for an international body — especially one with enforcement power over private AI labs — remains to be seen. The letter’s demands are considerably more specific than most policy frameworks currently under discussion, and considerably more intrusive for the companies developing these systems.