Mark Zuckerberg urges Trump to reject national AI regulatory framework

The push against a centralized regulator highlights a growing rift between tech giants over how to balance safety with speed. If the administration adopts Zuckerberg's light-touch recommendation, it could accelerate the deployment of generative AI tools across streaming and media platforms by removing federal pre-clearance hurdles. This stance directly challenges the safety-first approach favored by researchers like Demis Hassabis, potentially setting a precedent for industry-led self-regulation. The outcome will likely dictate the pace at which Meta and its peers integrate frontier models into consumer-facing video applications. Watch for the specific policy leanings of upcoming White House appointees to the proposed regulatory body as a signal of which faction has won the administration's favor.

Meta has been the most vocal opponent of mandatory federal AI safety testing among major platform companies. In July 2025, Meta published a policy paper arguing that existing U.S. regulatory agencies already possess sufficient authority to oversee AI without new legislation, a position that aligns with Zuckerberg's reported lobbying on the private call. The company's stance places it in direct opposition to Google DeepMind, whose CEO Demis Hassabis has repeatedly called for international safety standards on frontier models. In a May 2025 interview, Hassabis told the Financial Times that voluntary commitments alone cannot prevent catastrophic misuse of advanced AI systems, reinforcing the split between Meta's deregulatory posture and DeepMind's safety-first framework. The White House has signaled openness to industry input while stopping short of endorsing any single proposal. In June 2025, David Sacks, the administration's AI and crypto czar, outlined a framework that would create a FINRA-style self-regulatory organization for frontier AI developers, requiring companies to submit safety cases before deploying models above a defined compute threshold. The proposal drew immediate pushback from Meta and xAI, both of which argued that pre-market approval would slow U.S. model development relative to Chinese competitors. Elon Musk, whose xAI operates the Grok model, publicly criticized the Sacks framework on X in June 2025, calling it a regulatory capture mechanism that would entrench incumbents. The tension between Musk and Sacks, both Trump allies, underscores how the administration's own coalition is fractured on the question of federal AI oversight. The regulatory debate carries direct implications for how quickly generative AI tools reach streaming and media applications. Meta's Llama 4 models, released in April 2025, were made available under an open-weight license specifically to avoid the compliance burden that a pre-market testing regime would impose, a strategic choice that would become harder to sustain under a FINRA-style framework. Meanwhile, Google DeepMind's Gemini 2.5 Pro model, released in March 2025, included built-in safety evaluations that the company says would satisfy most proposed pre-deployment testing requirements, suggesting that firms already investing heavily in safety infrastructure would face lower compliance costs under any new regime. The outcome of this policy fight will determine whether streaming platforms integrating frontier models face a uniform federal approval process or continue operating under a patchwork of voluntary commitments.