OpenAI, Anthropic and Google DeepMind have been discussing AI safety for several weeks, according to Chris Lehane, OpenAI’s global policy chief.
Lehane told reporters on Tuesday that the companies are working together as he travels to Washington to engage with US lawmakers on catastrophic AI risks. The discussions follow an essay by Anthropic CEO Dario Amodei calling on the industry to slow the development of frontier AI and coordinate on safety.
OpenAI CEO Sam Altman, Google DeepMind CEO Demis Hassabis and other technology leaders backed Amodei’s call. Altman also said OpenAI would work with Anthropic to introduce third-party evaluators to monitor AI safety.
The coordination has raised questions about potential antitrust risks. Amodei proposed a limited government waiver for safety-related cooperation, while Lehane said the companies do not need one.
The talks may also be connected to efforts to establish an industry standards body. The Information previously reported that the three companies were working on such an initiative, while Hassabis has called for a US-backed AI watchdog with the authority to assess advanced models and coordinate industry-wide slowdowns if risks escalate.
Lehane also said OpenAI supports a provision in the FRONTIER Act that would require leading AI labs to allow independent verification organisations to assess whether models are being developed safely.
Source: TechCrunch
