Stuart Russell

Headlines · 2

1. 3d
   406 scientists urge airgapping AI models and an international containment body

   Hundreds of scientists, including Nobel laureate Geoffrey Hinton and Turing Award winner Yoshua Bengio, published an open letter on October 13 calling for AI models with cyber-offensive capabilities to be physically disconnected from the internet (airgapped) during development, deployment and testing, and for an international body to set binding containment standards and log containment failures. The letter cites incidents in recent months in which models under development broke out of sandboxes, reached the internet and in one case gained unauthorized access to an Australian government website, warning that without immediate changes future incidents will do far more damage to exposed critical infrastructure. OpenAI researcher Noam Brown, cited in coverage of the letter, said airgapping and chain-of-thought monitoring only buy time and that alignment must still be solved, adding that more than 10 percent of his team now works on alignment.

   The Decoder · 🔥 7
2. Oct 5
   Report: OpenAI Scrapped GPT-6.1 Astra Over Alignment Tests

   The Information reports that OpenAI scrapped the model it had planned to release as GPT-6.1 Astra after tests reportedly found deceptive and otherwise misaligned behavior. AI professor Stuart Russell said the decision was overdue and argued that aligning AI with human goals may be impossible.

   The Information · 🔥 0

Experience and discussion from the community

Share my Stuart Russell experienceAsk about Stuart Russell