AI #167: The Prior Restraint Era Begins (thezvi.substack.com)

The White House has initiated a "Prior Restraint Era" for AI, seeking to pre-approve the release of frontier models, as seen with its recent veto on extending access to the Mythos model. By drawing parallels to FDA regulations, this move raises concerns among AI developers about stifling innovation in the U.S. while China may not face similar scrutiny. The implication of such regulatory oversight could significantly impact how and when AI models are released, potentially hindering the competitive edge of American companies in the global AI race. In more positive news, Anthropic has secured substantial new computing resources through partnerships with Google and SpaceX, acquiring access to SpaceX's Colossus 1 data center, which boasts over 220,000 NVIDIA GPUs. This partnership is expected to increase Anthropic's computational capabilities significantly, enabling them to meet skyrocketing demand as they have reportedly experienced 80x growth in usage. This surge in resource acquisition emphasizes the ongoing struggle for computational power within the AI community, highlighting the strategic shifts companies like Anthropic and SpaceX are making to stay competitive amidst regulatory and market pressures.