본문으로 건너뛰기
All news

White House Set to Publish Voluntary Frontier AI Model Standards This Week

Summary: The White House is poised to release voluntary AI model launch standards this week — a concrete follow-through on Trump's June 2 executive order — establishing 30-day pre-release federal review windows for frontier AI models with advanced capabilities.

Key Points

  • Follows Trump's June 2 EO "Promoting Advanced AI Innovation and Security," which created a "Secure Frontier Model Deployment" framework
  • Core mechanism: models with advanced cyber capabilities reviewed by NSA and other federal agencies for up to 30 days before broad commercial release
  • GPT-5.6 Sol's general availability is reported to depend on satisfying these benchmarks
  • Anthropic confirmed willingness to develop joint evaluation standards alongside Amazon, Microsoft, and Google
  • Explicit limit: "nothing shall authorize mandatory licensing, preclearance, or permitting requirements" — strictly voluntary cooperation

Why It Matters

This is the first time U.S. AI policy has moved from broad safety language to structured, timed checkpoints. Even on a voluntary basis, federal review windows and benchmark requirements risk hardening into de facto industry practice — reshaping launch timelines for future Claude and GPT releases and creating a policy template that other governments will study and adapt.

Read More

뉴스레터 구독

무료 뉴스레터

매주 핵심 AI 소식, 한 번에 받기

쏟아지는 AI·LLM 뉴스 중 꼭 알아야 할 것만 골라 메일로 보내드려요. 뉴스레터 발송이 시작되면 구독자분들께 가장 먼저 보내드립니다.