Trump admin nears AI standards deal with Big Tech

The Trump administration is reportedly finalizing a set of voluntary cybersecurity standards for frontier AI models, with an announcement expected as early as next week. According to the *Financial Times*, the deal involves major U.S. AI companies including Anthropic, OpenAI, Amazon, Microsoft, and Google. The standards will focus on assessing cyber capabilities of AI models, with the National Security Agency (NSA) and the Commerce Department's Center for AI Standards and Innovation (CAISI) playing central roles in formalizing these benchmarks.

This move marks a significant shift from the administration's earlier laissez-faire approach to AI regulation. Recent actions, including an export control directive that temporarily blocked Anthropic's latest model and an executive order mandating classified benchmarking processes, signal a more hands-on regulatory stance. The classified nature of these benchmarks means the public won't know the exact standards, but shared practices across companies will likely reveal parts of the agreement. Notably, Meta has reportedly resisted joining the deal, creating a potential gap in industry-wide adoption.