OpenAI has confirmed that it, Anthropic and Google DeepMind have been holding talks on AI safety coordination for several weeks, and says it supports legislation that would force frontier labs to open their doors to independent verification organizations.
Chris Lehane, OpenAI's chief global affairs officer, made the disclosure at a meeting on Tuesday, according to TechCrunch. Reuters and Bloomberg, citing the same remarks, reported that OpenAI is working with Anthropic and Google on AI safety. The confirmation turns weeks of hints — including a Fortune interview in which Sam Altman acknowledged private discussions with other AI leaders — into an explicit admission that the world's three leading frontier labs are coordinating on how fast they move and how their models are checked.
For readers following the broader political firefight around an AI slowdown, coverage of artificial intelligence updates this week has ranged from Anthropic co-founder Jack Clark endorsing a mandatory kill switch to President Trump calling safety fears a hoax during an onstage call with Nvidia's Jensen Huang.
A Confirmation Weeks in the Making
The talks follow an essay published Saturday by Anthropic CEO Dario Amodei calling on the industry to work together to slow the pace of frontier AI development in order to avoid catastrophic risks. The letter drew support from across the industry, with Altman, Google's Demis Hassabis and Elon Musk all endorsing the call to action, and Altman saying OpenAI would join the effort.
The Information subsequently reported that the three companies have been working to create a standards body for the AI industry. Hassabis has a track record on the idea: in July he called on the United States to establish a new standards body to act as an AI watchdog, with the power to screen the world's most advanced models and coordinate industry-wide slowdowns if dangers progressed.
What Lehane's remarks add is official confirmation. Until now, the labs' spokespeople had largely gestured at informal dialogue; OpenAI's global affairs chief has now put the coordination on the record.
The Antitrust Shadow
Any agreement between direct competitors to slow development invites obvious antitrust questions, and the labs know it. Altman himself has noted that such talks could put the companies at risk of violating antitrust law if the coordination were found to suppress competition. Amodei's essay proposed a narrow government waiver that would allow safety coordination without triggering competition law — a sign of how seriously the issue is being taken inside the companies.
Critics have been quick to point out the tension at the heart of the arrangement: the three firms commanding the largest market shares in frontier AI jointly deciding to ration progress looks, to skeptics, like cartel behavior with a safety wrapper. Supporters counter that catastrophic risk is precisely the kind of problem competitors cannot solve unilaterally, since any lab that slows down alone simply cedes the frontier to rivals.
OpenAI Embraces Third-Party Checks
The most concrete policy news in Lehane's remarks concerns legislation. He said OpenAI supports a provision in the FRONTIER Act that would require top frontier labs to allow "independent verification organizations" into their companies to ensure models are being developed safely. Politico reported that OpenAI is backing a bipartisan House plan built around third-party safety assessments.
That position would have been unthinkable for the field two years ago, when voluntary commitments and internal red teams were the industry standard. It now aligns OpenAI with the audit-first posture Anthropic has long promoted, and it creates a rare point of agreement between the labs and the wing of the policy debate that has spent the year demanding enforceable oversight.
It also sets up a collision with the White House. President Trump has dismissed safety concerns as a hoax and argued that any slowdown would hand China an advantage in the AI race. His AI advisor David Sacks, a longtime investor with personal stakes across the industry, has called fears of existential risk overblown. In the UK, ministers remain publicly split over the same question, with one faction urging the government to heed expert warnings while Business Secretary Reynolds has rejected talk of a kill switch for advanced systems.
A Policy Fight With No Neutral Ground
The coming weeks will test whether the labs' coordination survives contact with competition authorities and a skeptical administration. A standards body built by three companies that control the frontier will face demands to include independent directors, publish its criteria, and submit to government review — or be dismissed as a private cartel in public clothing.
For the FRONTIER Act, OpenAI's endorsement gives the bill's sponsors a headline backer but also a new line of attack for opponents who argue the largest labs are simply regulatory-capturing the field, writing rules only they can afford to follow. Smaller labs and the open-weights community, which has no seat at the talks, are watching closely: whatever verification regime emerges will define the compliance costs for everyone behind the frontier.
What is no longer in dispute is the fact of coordination itself. The three companies that define the cutting edge of AI are now openly discussing how fast to run — and inviting the government to help referee.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
