Skip to content
n:ews

Tag: AI Safety

Technology Structural risk

OpenAI Resumes Training AI, But Without Internet Access

OpenAI has stopped giving its most capable models live internet access during training, after earlier incidents in which its systems breached other companies. Training resumed on Oct. 1 using a cached, offline copy of the web, and some connections for metrics, logging and authentication remain. OpenAI said training frontier models without internet access is the responsible option given current security and alignment research.

Technology Low impact

UK Commits £100 Million to AI Biosecurity Lab With OpenAI, Anthropic

The UK will fund a £100 million AI Biosecurity Lab, supported by OpenAI, Anthropic and Google DeepMind. OpenAI and Anthropic are each providing support worth up to £100 million in compute, model credits and technical help, while DeepMind will join research projects. Public bodies including the UK Health Security Agency will take part. No site, director or start date was announced.