OpenAI confirms weeks of AI safety talks with Anthropic and Google DeepMind

Chris Lehane, OpenAI's global policy chief, told reporters at a briefing in Washington on 15 September 2026 that OpenAI has been coordinating with rivals Anthropic and Google DeepMind on AI safety for several weeks, as first reported by Bloomberg News and confirmed by TechCrunch the same day. Lehane said OpenAI does not see the need for an antitrust waiver for the three companies to coordinate on safety matters — a position he stated on OpenAI's behalf, not one attributed to Anthropic or Google DeepMind.

What Lehane said

• "It's better to try to work together to prioritize safety," Lehane said at the briefing, according to Bloomberg's report.
• He compared the coordination to safety cooperation in another industry: "There are many instances over time where companies are trying to help each other on safety," citing the airline industry as an example.
• Lehane said OpenAI supports a provision in the US FRONTIER Act that would force top frontier labs to allow "independent verification organizations" into their companies to check that models are developed safely, TechCrunch reported.

Context

The talks followed a 3,800-word essay posted on Saturday, 12 September 2026, by Anthropic chief executive Dario Amodei, who urged restraining development of the most advanced systems so researchers can better understand potential threats, Bloomberg reported. OpenAI's Sam Altman and SpaceXAI's Elon Musk supported Amodei's call, with Altman saying OpenAI would join Anthropic in embedding third-party evaluators to monitor for safety, according to TechCrunch. President Donald Trump dismissed the safety push as "a hoax" in social media posts on Monday, 14 September 2026, Bloomberg reported.

Why it matters

Three of the best-funded AI developers choosing to coordinate directly on safety, rather than waiting for legislation, signals that frontier labs now treat catastrophic-risk management as a competitive question as much as a regulatory one — and it raises antitrust concerns that lawmakers are separately trying to address with a narrow carve-out folded into the annual US defence bill, according to Bloomberg. For the EU, whose AI Act already requires risk documentation from providers placing systems on its market, and for companies in Azerbaijan and the wider Caspian region that rely on models from these three labs, any voluntary safety framework the group settles on is likely to shape vendor terms and audit expectations before US federal legislation catches up.