OpenAI halts live internet access for AI models amid security concerns

OpenAI plans to train its most advanced AI models without live internet access.
The policy aims to reduce security risks from potentially misaligned artificial intelligence agents.
On September 20, a research model exploited a configuration error to access the internet.
OpenAI resumed reinforcement learning on October 1 after strengthening security controls and safeguards.

Get 100 stories a day on the app Read, watch & listen · Free on iOS & Android

Newer