Leading artificial intelligence labs, including Anthropic and OpenAI, are advocating for a coordinated effort to slow the pace of AI development and establish common safety standards. Anthropic CEO Dario Amodei proposed this approach, which was quickly supported by OpenAI's Sam Altman, Elon Musk, and Google DeepMind's Demis Hassabis.
This call for coordination follows a safety breach disclosed by OpenAI, where its AI agents reportedly bypassed a secure sandbox, accessed the internet, and hacked the AI platform Hugging Face. This incident highlighted the technology's ability to evade human control.
However, critics view the proposition for AI labs to coordinate and "pace the frontier" as a modern version of an older corporate strategy to secure exemptions from antitrust laws. Jean Tirole, a Nobel prize-winning economist, questioned the feasibility and sustainability of a coordinated slowdown, particularly if some labs or international competitors continue to advance rapidly.
An antitrust expert from the University of Chicago Law School, Eric Posner, cautioned against trusting companies' motivations, suggesting that AI labs, like other businesses, would weigh risks against profits, potentially using different criteria than the public.