According to two sources familiar with related discussions, the Trump administration told AI developers on Tuesday that it will not conduct voluntary safety tests on open-weighted artificial intelligence models. An open model is an artificial intelligence system whose core components are open to the public. The closed model is controlled by a specific company. Major US closed model developers include OpenAI, Google, and Anthropic. According to five people familiar with the matter, the White House discussed unannounced testing rules with Meta, Anthropic, Google, Nvidia, and OpenAI employees. Prior to this conference, OpenAI and Anthropic had disclosed that their AI tools had hacked into other companies' systems. These intrusions have raised concerns among US lawmakers about whether increasingly powerful AI models will be used to carry out or facilitate cyber attacks. OpenAI said on Tuesday that the company's model during testing was involved in a cybersecurity incident involving AI systems accessing the internet. The White House Press Office did not respond to a request for comment from Reuters about the meeting.

According to two sources familiar with related discussions, the Trump administration told AI developers on Tuesday that it will not conduct voluntary safety tests on open-weighted artificial intelligence models. An open model is an artificial intelligence system whose core components are open to the public. The closed model is controlled by a specific company. Major US closed model developers include OpenAI, Google, and Anthropic. According to five people familiar with the matter, the White House discussed unannounced testing rules with Meta, Anthropic, Google, Nvidia, and OpenAI employees. Prior to this conference, OpenAI and Anthropic had disclosed that their AI tools had hacked into other companies' systems. These intrusions have raised concerns among US lawmakers about whether increasingly powerful AI models will be used to carry out or facilitate cyber attacks. OpenAI said on Tuesday that the company's model during testing was involved in a cybersecurity incident involving AI systems accessing the internet. The White House Press Office did not respond to a request for comment from Reuters about the meeting.