Sacks Rejects Regulatory Frameworks for AI Frontier
AI-generated image · US National Wire
David Sacks argues that OpenAI and Anthropic are pursuing regulatory capture under the guise of safety.
David Sacks has pushed back against calls for a formal regulatory approval process to manage the development of artificial intelligence, characterizing such efforts as potential "regulatory capture," as he first reported via Hacker News.
Writing on X, Sacks responded to comments from Dario Amodei and Sam Altman regarding the need to "pace the frontier." Sacks asserts that OpenAI and Anthropic currently maintain a duopoly on frontier intelligence based on revenue growth, model capability, and market share. While Sacks stated he supports the decision to slow development if unreleased models are deemed dangerous, he argued that the companies do not require external permission or a regulatory framework that supersedes product liability to do so.
Sacks contends that the drive toward "alignment" is rooted in business necessity rather than pure altruism, noting that the market penalizes unpredictable models and that the companies face significant product-liability exposure if their tools enable damaging cyberattacks. He specifically questioned the independence of METR, claiming the organization is intertwined with Anthropic's staff and investors.
Furthermore, Sacks warned that demanding a specific regulatory framework in exchange for slowing development would appear as "blackmail" to the political system and the public. He also noted that China is unlikely to participate in a global agreement, which should be a factor in the conversation. Sacks concluded that if the leaders of the frontier wish to avoid building superintelligence, they can simply agree to stop without seeking a preferred regulatory structure.