Source: before the Hugging Face incident, OpenAI was negotiating a legally binding deal with Anthropic for the companies to stress-test each other's models (The Information)

Before the Hugging Face incident, OpenAI was reportedly negotiating a legally binding deal with Anthropic. The purpose of this agreement was for the companies to stress-test each other's AI models. This negotiation comes as OpenAI rethinks its safety strategies in response to concerns from employees and others regarding the potential dangers of its artificial intelligence.

This report reveals a previously unknown, legally binding negotiation between OpenAI and Anthropic to stress-test models, unlike their current informal safety collaborations.