Trump rejects formal AI licensing, opts for ad hoc emergency powers

In a recent interview, outgoing White House AI advisor Sriram Krishnan explicitly stated that President Trump will never support a formal licensing regime for artificial intelligence. Instead, the administration will continue its current approach: using emergency powers to stall the most advanced models on an ad hoc basis. Krishnan argued that centralized agencies requiring 'a team of lawyers before you can get a model out' would hinder innovation, and that 'we are not in the business of picking winners and losers.' However, critics point out that the opaque, arbitrary process effectively does pick winners and losers based on the whims of the administration.

Krishnan also defended the use of export controls and admitted having no concern for how future governments might misuse these powers. The delay of GPT-5.6 by a few weeks was cited as a minor inconvenience, but many worry about far worse outcomes. The article further discusses the implications of this ad hoc regime, including the potential for government overreach, the lack of transparency, and the risks to AI safety and alignment research. The discussion is part of a broader analysis of AI policy, including critiques of open-weight models and Nvidia's compliance with government requests.