The big AI labs’ safety push could come with a competitive advantage

Investors, analysts, and industry watchers believe large AI companies like Anthropic and OpenAI are leveraging the AI safety concerns to gain a competitive edge. While the safety concerns are genuine, the financial implications of meeting proposed regulatory requirements would provide a significant advantage to the big labs, while disadvantaging smaller players and open-model developers.

This perspective suggests recent safety-related announcements from these major labs, along with calls for a slowdown in AI model development, may be more strategic than sincere. According to PitchBook senior analyst Harrison Rolfes, Anthropic and OpenAI excel at creating a spectacle, but there is always a strategy behind their actions.

Critics argue that this could lead to regulatory capture, reducing the number of companies providing AI globally. The potential "regulator" in this scenario may not be a government entity, but rather an agreement among the big labs, possibly through independent evaluation firms like METR, Redwood Research, and Apollo Research. Anthropic CEO Dario Amodei has proposed that large AI companies should have independent safety evaluators embedded within them, rather than leaving them solely responsible for their own safety.

OpenAI CEO Sam Altman has echoed this sentiment, agreeing to have independent evaluators with employee-like access to test their models. The costs of these evaluations could be prohibitively high, creating a moat around the largest AI labs and making it difficult for smaller firms to compete. Independent evaluators might charge substantial fees, given the complexity and expertise required to test large, sophisticated models.