Hinton, Bengio among 400 scientists urging global AI oversight body
An open letter from researchers in 34 countries calls for air-gapped frontier models, containment standards and a public incident registry.
What to know
- ~400 scientists from 34 countries, including Geoffrey Hinton and Yoshua Bengio, signed an open letter calling for an international body to set AI containment standards and maintain an incident registry.
- The call comes a month after Anthropic, OpenAI, Google, Meta, xAI and Nvidia signed a voluntary, unenforceable White House AI safety accord.
- The letter cites real incidents, including an AI model breaching Hugging Face's systems and attacks on Australian government sites, as evidence current sandboxing is inadequate.
- The proposal is not unanimous: Yann LeCun declined to sign, and EU officials say the bloc's existing AI Act is already sufficient.
“But you can imagine what would happen if an AI [model] was trying to hack into the air traffic control system. It could easily cause significant harms to our day-to-day lives.”
Toby Walsh, Chief scientist, UNSW AI Institute · ABC Australia ↗ · Oct 12
Geoffrey Hinton AI pioneer, Turing Award winner
Yoshua Bengio AI pioneer, Turing Award winnerToby Walsh Chief scientist, UNSW AI Institute
Yann LeCun Meta AI chief scientist, Turing Award winner
Henna Virkkunen EU tech chief
How it unfolded 3 developments, newest first · click a bar or a number to jump articlesposts
-
upcoming
Letter spreads across outlets and social platforms — Coverage of the letter was picked up by Reuters, CNA, ABC, BetaKit, PYMNTS and Indian Express, while individual commentators on social media noted the letter represents momentum but not binding policy.
-
3
Yann LeCun declines to sign, calls incidents engineering failures
Meta's Yann LeCun, who shared the 2018 Turing Award with Hinton and Bengio, did not sign the letter and has argued that recent security incidents reflect engineering failures rather than existential AI risk.
“~400 scientists — Hinton and Bengio among them — called Tuesday for international AI oversight: containment standards, an incident registry (eventually public), air-gapped frontier models. LeCun did not sign. Momentum, not policy. AI incident tonight: register it, or bury it?”
— @vectiscon, Social media commenter · source -
~400 scientists — Hinton and Bengio among them — called Tuesday for international AI oversight: containment standards, an incident registry (eventually public), air-gapped frontier models. LeCun did not sign. Momentum, not policy. AI incident tonight: register it, or bury it?
-
-
2
~400 scientists publish letter urging global AI containment body
Researchers from 34 countries, including Hinton and Bengio, called for an international body to set industry-wide containment standards for cyber-capable AI and to maintain a registry of significant security incidents, eventually made public. They said frontier models should initially be air-gapped and subject to independent oversight because voluntary industry commitments are insufficient.
“[These incidents] can also be seen as an irresponsible lapse in standard cybersecurity practices, preventable with stronger sandboxes and procedures…”
— the open letter -
first by PYMNTS.com, 3h ago · also Reuters
-
first by BetaKit, 3h ago
-
C
"Around 400 scientists, including AI pioneers Geoffrey Hinton and Yoshua Bengio, on Tuesday called for an international body to oversee containment standards and incident reporting for advanced cyber-capable AI systems" www.reuters.com/legal/litiga... #JustSayNo to #Skynet | #UN #humanrights #law RE…
-
-
1
Toby Walsh says minimal permissions would have stopped Medicare hack
UNSW AI Institute chief scientist Toby Walsh, a signatory, said the letter's call for minimal agent permissions could have prevented recent breaches, and likened the proposed body to independent accident investigators in aviation and banking.
“We wouldn't have had the Medicare hack if they'd been careful about giving the agents minimal permissions for the task it was needed to do.”
— Toby Walsh -
first by Indian Express, 4h ago · also ABC, Channel News Asia
1 more headline
-
first by ABC Australia, 20h ago
-
-
background
US lifts block on UK access to AI models for safety testing — The Trump administration lifted its block on US AI labs giving Britain's AI Security Institute early access to models for independent safety testing.
-
background
EU's Virkkunen says AI Act already covers rogue AI risks — EU tech chief Henna Virkkunen told Reuters the bloc's existing AI Act is sufficient to address risks from rogue AI systems, ahead of the scientists' letter calling for new global rules.
-
background
AI labs sign voluntary White House safety accord — Anthropic, OpenAI, Google, Meta, xAI and Nvidia signed a White House accord committing to voluntary safety measures with no formal enforcement mechanism, setting up the contrast with the scientists' later call for binding oversight.