Editorial illustration for AI Firms in Talks Over Safety for Weeks
OpenAI, Anthropic, Google DeepMind Discuss AI Safety
AI Firms in Talks Over Safety for Weeks
Chris Lehane, OpenAI's global policy chief, told reporters Tuesday that his company has spent the past several weeks talking with Anthropic and Google DeepMind about AI safety, confirming a report first published by Bloomberg. Lehane made the comments while in Washington, where he's been meeting with lawmakers about catastrophic risks tied to advanced AI systems.
The disclosure lands days after Anthropic CEO Dario Amodei published an essay Saturday urging the industry to coordinate on slowing frontier AI development to avoid worst-case outcomes. Sam Altman, Demis Hassabis and Elon Musk all voiced support for Amodei's call, with Altman saying OpenAI would follow Anthropic's lead in bringing outside evaluators into the company to check for safety problems.
That kind of cooperation between competitors raises an obvious legal question. Altman himself has pointed out that three of the biggest names in AI comparing notes could run into antitrust trouble if regulators decide it amounts to suppressing competition rather than managing risk. Amodei's essay floated a narrow government waiver to cover exactly this kind of coordination.
On Tuesday, Lehane admitted that the three companies have been in talks for several weeks regarding AI safety. Some, including Altman, have noted that such talks could put the companies at risk of violating antitrust law if the coordination is found to suppress competition.
Why this matters
Three companies that spend most of their time suing each other over talent and compute are suddenly comparing notes on catastrophic risk, and that alone is worth watching closely. Lehane's trip to Washington isn't a coincidence: safety cooperation announced right as lawmakers are drafting AI rules gives OpenAI, Anthropic, and Google DeepMind a chance to shape regulation before it's written, not after. For developers and founders building on these platforms, third-party evaluators embedded inside labs could mean real scrutiny of model behavior, or it could mean a managed process designed to satisfy regulators without slowing anyone down much.
Amodei's essay asked the industry to slow frontier development; Altman's response was to agree to oversight while presumably continuing to ship. Those aren't the same thing. Researchers should ask who these evaluators report to, what triggers a pause, and whether findings get published or stay internal.
Coordination among competitors on safety is rare enough to notice. Whether it changes what actually ships, on what timeline, is the part still unproven.
Related Reading
Policy & Regulation Akamai Proposes SAFE Cybersecurity Guidelines for AI Threats
Policy & Regulation After ditching AI fitness apps and a Fitbit, I return to Peloton classes
Policy & Regulation Sony Music, Warner Chappell Sue Anthropic Over "Blatant Theft
From Archives OpenAI Funds Biological Data Collection to Aid AI Disease Research
From Archives AI Assistant Kit Aims to Reclaim Users' Time
Common Questions Answered
Which three AI companies have been in talks about AI safety for several weeks?
OpenAI, Anthropic, and Google DeepMind have been engaged in discussions regarding AI safety, according to Chris Lehane, OpenAI's global policy chief. Lehane confirmed these talks while meeting with lawmakers in Washington to discuss catastrophic risks associated with advanced AI systems.
What antitrust concerns have been raised about the AI safety coordination between these companies?
Some industry figures, including Sam Altman, have noted that coordination between OpenAI, Anthropic, and Google DeepMind on AI safety could potentially violate antitrust law if the discussions are found to suppress competition. The coordination between these competing companies raises questions about whether their safety discussions cross into illegal territory.
Why is the timing of these AI safety talks significant for regulatory development?
The announcement of safety cooperation comes at a critical moment when lawmakers are actively drafting AI regulations, allowing OpenAI, Anthropic, and Google DeepMind to shape regulatory rules before they are finalized rather than after implementation. This strategic timing gives these companies significant influence over how AI safety requirements will be codified into law.
What did Anthropic CEO Dario Amodei advocate for in his recent essay?
Anthropic CEO Dario Amodei published an essay urging the AI industry to coordinate on slowing the development of advanced AI systems. His essay appears to align with and provide context for the ongoing safety discussions between the three major AI companies.
Further Reading
- Anthropic, OpenAI and Google held talks on AI safety body ... - InvestingLive
- OpenAI, Anthropic, and Google in talks to enhance AI safety - CryptoBriefing
- Anthropic, OpenAI, and Google DeepMind in talks to form ... - CryptoBriefing
- OpenAI, Anthropic, Google to Join White House AI Safety ... - Bloomberg
- OpenAI makes U-turn and calls for binding national AI safety rules - Euronews
