OpenAI, Anthropic, Google Confirm Weeks of AI Safety Talks — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Chris Lehane, OpenAI's global policy chief, confirmed in Washington on Tuesday that OpenAI, Anthropic, and Google DeepMind have been working together on AI safety for several weeks
- Anthropic CEO Dario Amodei published an essay Saturday calling on the industry to slow frontier AI development and avoid catastrophic risks, a call supported by Sam Altman, Demis Hassabis, and Elon Musk
- Sam Altman said OpenAI would join Anthropic in embedding third-party evaluators into the company to monitor for safety; Altman flagged antitrust risk, Amodei proposed a narrow government waiver, and Lehane said the firms don't need one
- Lehane said OpenAI supports a FRONTIER Act provision requiring top frontier labs to allow "independent verification organizations" in to vet that models are developed safely
- The three companies are working together to create an AI industry standards body, per The Information, with Hassabis having called on the U.S. in July to establish a watchdog with power to screen advanced models and coordinate industrywide slowdowns
- President Trump dismissed AI safety concerns as a hoax and opposed regulation, arguing any slowdown would give China a leg up; his AI advisor David Sacks said existential risk fears are overblown
Why it matters: Three frontier AI labs are already coordinating safety measures on their own terms, without requesting a government waiver, putting pressure on Congress to pass the FRONTIER Act's verification provision. The coordination sets up a direct clash: industry wants self-regulation with embedded evaluators, while the White House actively frames any slowdown as ceding ground to China.
Ask SkimNews
