OpenAI’s Lehane Says Safety Talks With Anthropic and Google Are Under Way
On September 15, 2026, OpenAI global policy chief Chris Lehane told reporters in Washington that talks with Anthropic and Google DeepMind on AI safety have run for several weeks and do not need an antitrust waiver. Same day, Jensen Huang and Mark Zuckerberg rejected a coordinated slowdown.
TLDR
Bloomberg, corroborated by Reuters and TechCrunch, on Tuesday, September 15, 2026: Chris Lehane, OpenAI’s global policy chief, told a Washington briefing that engagement with Anthropic and Google DeepMind on AI safety has been under way for several weeks. Quote: “It’s better to try to work together to prioritize safety.” He said the three firms do not need an antitrust waiver to coordinate on safety—contra the waiver Dario Amodei floated in “Pace the Frontier” (Sep 12). Lehane is in DC to meet lawmakers; OpenAI would support bipartisan catastrophic-risk bills (“whatever we can get through”), including the Obernolte–Trahan federal-governance framework and a Senate track associated with Thune, Cruz, and Klobuchar. TechCrunch: he also backs a FRONTIER Act provision for independent verification organizations inside labs.
Same-day counter, do not dual-file Sep 16: Jensen Huang at Dreamforce and on CNBC Mad Money—safety is an engineering problem; “we don’t need any new laws”; an antitrust waiver is “completely unnecessary”; “run as fast as you can,” pause only if the product is unsafe. Mark Zuckerberg on X (Tue): labs already have incentive and liability; Meta delayed Muse for months without asking others to pause; independent evaluators are “industry best practice.”
Product-line de-dupe: not Amodei’s Sep 12 essay (already filed), not the Sep 13 standards-body talks (re-pass, unfiled), not Coxon (Sep 9). New fact: Lehane on the record that the talks are already happening and no waiver.
Why this story matters
Saturday’s coalition was a blog. Tuesday is OpenAI’s policy shop in Washington saying the coordination is live—while the chip vendor and Meta say don’t slow, don’t waive. Watch: badges for evaluators vs Huang’s market-forces line, not the adjectives.
Sources
- Reuters: OpenAI is working with Anthropic, Google on AI safety, Bloomberg reports (September 15, 2026)
- TechCrunch: OpenAI, Anthropic, Google have been in talks on AI safety for weeks (September 15, 2026)
- CNBC: Nvidia’s Huang rips Anthropic’s antitrust-waiver proposal (September 15, 2026)
- Reuters: Zuckerberg says AI labs have enough incentive to build safely (September 15, 2026)