More than 20 leading AI researchers warn that automated AI research poses extreme risks
20+ AI researchers including Hinton and Bengio warn: self-improving AI could compress years of research into months. The governance gap just got wider.
Why it matters
A cohort of credible AI researchers is publicly escalating warnings about automated AI research and recursive self-improvement—moving the 'intelligence explosion' debate from theory to near-term governance and safety concern. Enterprise buyers and operators need to understand the policy and risk-management implications as AI autonomy accelerates.
The key facts
10 to know20+ leading researchers including Geoffrey Hinton, Yoshua Bengio, Jakub Pachocki (OpenAI research lead)
Warning: AI systems could automate all AI research, compressing years into months
Core concern: 'intelligence explosion' from self-improving AI
Published 28 September 2026
Source: The Decoder (original reporting attribution required)
Signatories include Geoffrey Hinton, Yoshua Bengio, Jakub Pachocki (OpenAI research lead)
Central claim: AI systems could automate all AI research, compressing years into months
Framing: 'intelligence explosion' and recursive self-improvement risk
No date disclosed for when automation risk becomes acute; no technical evidence provided for the compression timeline
Publication: The Decoder, Sept 28, 2026
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: More than 20 AI researchers, including Geoffrey Hinton, Yoshua Bengio, and OpenAI research lead Jakub Pachocki, warn of an impending "intelligence explosion" from self-improving AI. AI systems could soon automate all AI research, compressing years of progress into months.