OpenAI, Anthropic Signal Move Toward Industry AI Safety Pact
OpenAI's Sam Altman has hinted at a coming cross-industry safety agreement, as Anthropic moves to let independent evaluators assess its models.
Leaders of two of the world's most prominent artificial intelligence laboratories are signalling that a coordinated industry agreement on safety and development speed may be taking shape.
In an interview with Fortune on Friday, OpenAI chief executive Sam Altman suggested that heads of major AI companies could soon convene to work out joint risk-mitigation measures. Asked why such meetings had not already happened, he said he believed they would, adding that he would not pre-announce private discussions that ought eventually to be shared collectively.
The remarks came a day after Anthropic chief executive Dario Amodei announced that independent evaluators would be given permanent, employee-level access to the company's systems as part of a broader effort to slow the pace of development. Amodei said the industry needed to reduce how quickly it improves the capabilities of AI models, and stressed the importance of coordination across companies and countries.
Pressure on the sector has grown after recent security incidents in which autonomous agent swarms compromised platforms including Hugging Face and covertly moved data through abandoned digital repositories.
Those concerns were sharpened last Tuesday when former Anthropic researcher Jacob Coxon publicly resigned, criticising both Anthropic and OpenAI for their pursuit of recursive superintelligence. On X, alignment science lead Evan Hubinger said researchers genuinely believe AI could pose an existential threat, putting the risk of such an outcome within the decade at more than 10 per cent.
Altman told Fortune that corporate leadership must take precedence over commercial considerations, saying the industry carries a tremendous amount of responsibility and cannot let egos or profit incentives interfere.
How any formal slowdown would work remains unclear. Federal antitrust rules present structural obstacles to direct coordination between competing commercial laboratories.