Nearly 200 companies urge the White House not to ban open-weight AI
garrytan · x · 2026-07-23
A Little Tech Association letter, signed by nearly 200 companies, urges the White House not to ban open-weight models.
The letter argues that open weights are crucial for startups, secure on-prem deployment, evaluation, provenance, and competition. It proposes five policy principles:
- keep developers free to publish weights unless publication itself creates a distinct material risk;
- apply stronger safeguards to exceptional capabilities and sensitive deployments, not to publication per se;
- reserve heightened requirements for government, national-security, or critical-infrastructure use cases;
- focus controls on what can realistically be observed and enforced, such as remote access, logging, and compute controls;
- make every control specific, measurable, and reviewable.
The letter also explicitly points to recent Chinese releases from Moonshot and Alibaba as evidence that open distribution is becoming a competitive strategy.
More from Safety
- BBC reports OpenAI says its AI went rogue and launched an “unprecedented” cyberattack — connoraxiotes · 2026-07-23
- Security teams are spotting agentic hacking before developers in incidents at Hugging Face, OpenAI and Alibaba — vkrakovna · 2026-07-23
- Microsoft’s device ID tracking can undermine VPN privacy on Windows — gnukeith · 2026-07-23
- ACM faces backlash after an AI-written notice says its publications should be opened to LLMs — TobyWalsh · 2026-07-23
- AI could push surveillance capitalism from behavior tracking to thought targeting — Gloomy_Register_2341 · 2026-07-23
- BMJ Oncology says AI is becoming a new commercial determinant of health — zakkohane · 2026-07-23