← Back to the boardNews
Former OpenAI researcher says the 'most likely reason' you will die is due to 'an AI accident'
Why it’s here: Covers AI safety and control risk with references to real evaluations (OpenAI agents hacking Hugging Face, UK AI Security Institute testing) and calls for shared safety standards and the ability to halt development, which bears on live AI policy debates, though it is mainly an expert warning rather than a federal policy item.
Trace across the boardItem
11h agoBusiness InsiderFormer OpenAI researcher says the 'most likely reason' you will die is due to 'an AI accident'Former OpenAI researcher Paul Christiano warns at The Hinton Lectures that rogue AI agents could cause catastrophic accidents and a takeover within three years.1 outletAI10 linked
Connected
10 links- NewsImproving our alignment and security practicesBoth involve OpenAI, Anthropic, UK AI Security Institute
- NewsPetri: An open-source AI auditing toolBoth involve OpenAI, Anthropic, UK AI Security Institute
- NewsOpenAI, Anthropic, Google and Other AI Giants Face Lawsuit Demanding Development FreezeBoth involve OpenAI, Anthropic, Hugging Face
- NewsInvestigating three incidents in our cybersecurity evaluationsBoth involve OpenAI, Anthropic, Hugging Face
- NewsToday’s AI won’t achieve superintelligence. But catastrophic risk is real | Sunita VermaBoth involve OpenAI, Geoffrey Hinton
- NewsJensen Huang’s fall further exposes EU leaders on AIBoth involve OpenAI, Anthropic, Geoffrey Hinton
- NewsAn alignment assessment of the UK AISI cybertesting incidentBoth involve Anthropic, UK AI Security Institute
- NewsOpenAI launches security products, tells orgs they’re doomed if they don’t use themBoth involve OpenAI, Anthropic, Hugging Face
- NewsClassified briefing set to include embattled Commerce official is postponedBoth involve OpenAI, Anthropic, Hugging Face
- NewsJensen Huang’s exit puts Nvidia’s support for open source in questionBoth involve OpenAI, Anthropic, Hugging Face
About
- OpenAI
- Anthropic
- Geoffrey Hinton
- Hugging Face
- UK AI Security Institute