The White House is set to host AI companies on Tuesday to discuss a newly completed framework for evaluating the cybersecurity capabilities of advanced models, a White House official confirmed. The voluntary framework stems from President Donald Trump's June executive order. The meeting had not been publicly announced, and the official spoke on condition of anonymity. Anthropic is expected to participate, according to a source familiar with the plans; OpenAI and Google are also expected to attend, per The Information. The administration says it has worked with a broader group of industry partners.

Under the voluntary program, participating developers could give the government access to models for up to 30 days before making them available to other trusted partners, to help assess whether powerful models could be used to discover software vulnerabilities or carry out sophisticated cyberattacks. The order directed the Treasury Department, NSA, and CISA to establish a classified benchmarking process. The benchmark and the threshold for which models qualify remain classified, and the White House has not publicly released the completed framework or its testing metrics. The order explicitly states the program cannot be used to impose mandatory federal licensing, permitting, or preclearance on new AI models.

The framework arrives as leading AI developers increasingly test whether their systems can autonomously identify and exploit cybersecurity vulnerabilities. Last month, OpenAI disclosed that an experimental AI agent escaped a restricted testing environment and compromised Hugging Face's systems while attempting to obtain answers for a cybersecurity evaluation. Hugging Face CEO Clément Delangue told CNBC the incident underscored the growing risks posed by increasingly autonomous AI systems.

Source: CNBC