White House won't release new AI safety framework publicly
What's the story
The White House has no plans to publicly release its new framework for assessing advanced artificial intelligence (AI) models, according to Axios. The voluntary framework is said to have global implications for AI security. However, the details will only be shared with companies involved in the process. This secrecy leaves other stakeholders such as policymakers, researchers, and US allies outside the process guessing about how the administration intends to implement one of its key AI policies.
Stakeholder discussions
Framework discussions amid industry challenges
On Tuesday, the White House held staff-level meetings with industry representatives to discuss the newly completed AI framework.
Companies that weren't invited to these discussions are still unaware of its contents.
The move comes as the industry faces high-profile cyber-attacks and rapid advancements by Chinese AI developers, reigniting debates over how to handle open-source models.
Access concerns
Uncertainty over 'trusted partners' and open-source models
The White House has not clarified which "trusted partners" will get early access to advanced models under the framework. This includes whether any foreign governments would qualify.
The discussions also touched on open-source models, but no further details were provided.
NVIDIA staff participated in these meetings, with CEO Jensen Huang meeting Commerce Secretary Howard Lutnick and President Donald Trump last week.
Framework details
Details on the voluntary framework
The voluntary framework, as per a June executive order, is designed to give AI developers a way to work with the government. This is to determine if their models fall under its purview.
The framework is also expected to set rules for government access to AI models up to 30 days before release, covering confidentiality, cybersecurity, insider risk, intellectual property, and non-disclosure requirements.