The Trump administration has completed a voluntary framework for evaluating advanced artificial-intelligence models and will convene industry leaders Tuesday to discuss its rollout, a White House official said. The plan, stemming from a June executive order, would invite companies to share “frontier” models with the government for up to 30 days before public release, aiming to bolster federal cybersecurity and assess potential risks without imposing binding rules. The White House framed the approach as pro-innovation and light-touch, seeking to avoid stifling development while addressing dual-use concerns about powerful systems that could expose software vulnerabilities. Axios first reported the move. The initiative parallels private efforts—such as Anthropic’s pre-release access to its Mythos model for select partners—to harden defenses, and underscores Washington’s preference for voluntary guardrails as it weighs longer-term regulation.
Related articles:
– NIST AI Risk Management Framework 1.0
– Frontier Model Forum launches to advance safe and responsible AI
