OpenAI, Anthropic, Google DeepMind in weeks-long AI safety talks

According to TechCrunch, OpenAI has confirmed that it has spent several weeks in safety-focused talks with Anthropic and Google DeepMind. The outlet reports the discussions center on artificial intelligence safety practices across the three labs, though the published account does not detail specific commitments or a timeline for any resulting agreement.

TechCrunch situates the talks against a political backdrop in which the Trump administration has downplayed AI safety concerns, instead emphasizing the need for the United States to maintain pace with China’s AI development. The report frames the labs’ coordination as occurring alongside, rather than in response to, that administration posture.

The story does not specify which officials at each company confirmed the talks, nor does it quote directly from any of the three labs beyond OpenAI’s acknowledgment that the discussions have taken place. No details are given regarding whether other major AI developers have been included in the conversations, or whether the talks are expected to produce a joint statement, shared technical standards, or another form of agreement.

TechCrunch’s report, published September 15, is the sole source for this account. Further particulars on the substance of the safety discussions, including their origin and any near-term outcomes, remain unreported at this time.