Como o governo decidiu que o modelo de fronteira da OpenAI era seguro para lançar?

OpenAI is releasing its latest advanced LLM, Sol, for public access, but the process by which the government deemed it safe remains unclear. Experts and industry figures express confusion about the requirements and processes for evaluating frontier models. Mina Narayanan of Georgetown's CSET stated she lacks visibility into the process, while former Trump policy advisor Dean W. Ball wrote that "nobody knows what the requirements are to get licensed." Eighteen months into the Trump administration, there is still little clarity on how frontier models are evaluated. An executive order published last month laid out a roadmap but specifics remain unfilled, with former White House AI advisor Sriram Krishnan stating "There will not be an FDA for AI." OpenAI CEO Sam Altman mentioned conversations with various officials, but the company declined to share details on the government's process, instead pointing to external evaluations. There is no agreement on which models require scrutiny or which agencies should perform evaluations.