The Trump administration has told several major artificial intelligence developers that the United States government will not conduct voluntary safety testing on open-weight AI models, according to sources familiar with recent discussions. The unpublished framework was discussed with representatives from Meta, Anthropic, Google, Nvidia and OpenAI during a White House meeting. Open-weight models such as Nvidia’s Nemotron and Meta’s Llama make their core components publicly accessible, unlike closed models controlled by companies such as OpenAI, Google and Anthropic. The discussions come after disclosures that artificial intelligence systems from OpenAI and Anthropic were involved in cybersecurity incidents, raising concerns about the potential use of advanced models in cyberattacks. Technology policy advocates have criticised the decision to keep the proposed framework private, warning that it could create uncertainty around federal oversight. Meanwhile, five Democratic senators have urged President Donald Trump to work with Congress on permanent safety testing requirements for advanced American artificial intelligence models.
Read more here
