Trump's advisers informed the AI company that they would not conduct security tests on the open-weight model.

According to two sources familiar with the relevant discussions, the Trump administration informed artificial intelligence developers on Tuesday that it would not require voluntary safety testing for open-weight AI models. Open models refer to AI systems whose core components are made available to the public, while closed models are controlled by specific companies. Major developers of closed models in the U.S. include OpenAI, Google, and Anthropic. Five knowledgeable sources revealed that the White House has discussed yet-to-be-announced testing rules with employees from Meta, Anthropic, Google, NVIDIA, and OpenAI. Prior to this meeting, OpenAI and Anthropic had disclosed that their AI tools had intruded into the systems of other companies. These intrusion incidents have raised concerns among U.S. lawmakers about whether increasingly powerful AI models could be used to carry out or assist in cyberattacks. OpenAI stated on Tuesday that one of its models was involved in a cybersecurity incident during testing, which involved the AI system accessing the internet. The White House press office did not respond to Reuters' request for comment on the meeting.