Three of the world's most powerful AI companies have been working together behind the scenes for weeks to address safety concerns, marking a significant shift toward industry-wide coordination on existential risk. OpenAI's global policy chief Chris Lehane confirmed on Tuesday that OpenAI, Anthropic, and Google DeepMind have been in ongoing discussions about AI safety, following Anthropic CEO Dario Amodei's recent essay calling for the industry to slow the pace of frontier AI development and work together to prevent catastrophic outcomes.

What's Actually Happening Behind Closed Doors?

The revelation comes after Amodei published an essay on Saturday outlining a framework for industry-wide safety coordination. His call to action resonated across the sector, with major AI leaders including OpenAI's Sam Altman, Google DeepMind's Demis Hassabis, and Elon Musk publicly supporting the initiative. Altman went further, announcing that OpenAI would join Anthropic in embedding third-party evaluators into their companies to monitor for safety risks. According to reporting from The Information, the three companies have been working together to create a standards body for the AI industry, something Altman reportedly told staff would need to happen without government support.

The scope of these talks extends beyond informal discussions. Hassabis had previously called on the U.S. government in July 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 industrywide slowdowns if dangers emerged. Now, it appears the companies are moving toward creating such a body themselves, potentially circumventing the need for government involvement.

Why Are Antitrust Concerns Suddenly a Problem?

The coordination between these three companies raises a thorny legal question. Some industry observers, including Altman himself, have noted that such talks could put the companies at risk of violating antitrust law if the coordination is found to suppress competition. Amodei's essay proposed a narrow government waiver allowing such safety coordination to proceed without legal risk, but Lehane reportedly said the firms don't believe they need one.

This legal gray area reflects a broader tension in AI governance. Companies want to work together on safety, but doing so in ways that might be perceived as collusion creates regulatory exposure. The question of whether safety coordination constitutes illegal antitrust activity remains unresolved, and the companies appear to be proceeding with the assumption that their efforts fall outside antitrust restrictions.

How Are Policymakers Responding to Industry Self-Regulation?

• Government Support: Lehane stated that OpenAI supports a provision in the FRONTIER Act that would force top frontier labs to allow "independent verification organizations" into their companies to ensure models are developed safely.
• Political Skepticism: President Trump has dismissed safety concerns as a hoax and pushed back on the need for tighter regulations, arguing that any slowdown would give China a competitive advantage in the AI race.
• Industry Advisor Pushback: David Sacks, Trump's key AI advisor and a longtime investor with personal stakes in much of the industry, echoed this skepticism, saying fears of existential risk from AI were overblown.

The political landscape presents a significant challenge to the companies' safety coordination efforts. While the industry is moving toward self-regulation and third-party oversight, the current administration appears skeptical of both the existential risk narrative and the need for regulatory intervention. This creates a situation where companies are essentially trying to establish safety standards in the absence of government support or mandate.

What Does This Mean for the Future of AI Governance?

The talks between OpenAI, Anthropic, and Google DeepMind represent a pivotal moment in AI governance. Rather than waiting for governments to establish rules, the companies are attempting to create industry standards from within. This approach has both advantages and risks. On one hand, it allows companies with deep technical expertise to shape safety practices. On the other hand, it raises questions about whether industry self-regulation can adequately address existential risks without government oversight and enforcement mechanisms.

"The company has been working with rivals Anthropic and Google DeepMind on AI safety for weeks," said Chris Lehane, OpenAI's global policy chief.

Chris Lehane, Global Policy Chief at OpenAI

The creation of an industry standards body, if it materializes, would represent a significant departure from how AI development has been governed to date. Historically, AI companies have operated largely independently, with minimal coordination on safety practices. A formal standards body would establish shared expectations around model evaluation, safety testing, and risk disclosure. Whether such a body would have real enforcement power or would function more as a voluntary coordination mechanism remains unclear.

The weeks-long nature of these talks also suggests that the companies have been moving carefully, likely consulting with legal advisors about antitrust implications and with policymakers about the political feasibility of their approach. The fact that Lehane is in Washington working with lawmakers indicates that the companies are trying to build political support for their safety coordination efforts, even as the Trump administration expresses skepticism about AI existential risks.

What remains to be seen is whether this industry-led approach will prove sufficient to address the catastrophic risks that Amodei and others have warned about, or whether it will eventually require more formal government involvement and enforcement. For now, the three companies are betting that they can coordinate on safety without triggering antitrust action, and that their efforts will be seen as a responsible approach to managing the risks of advanced AI systems.

" }