The Fragility of Chinese Open-Source AI and Export Controls
teortaxesTex · x · 2026-07-08
Commenter teortaxesTex highlights the hidden risks in the industry's reliance on "Chinese open-source models." Many assumed a fallback existed in models like GLM, which were thought to be around 8 months behind and constrained by Dutch export controls. However, Llama 3-405B was already technically outdated at launch and subsequently shelved, proving that open-source supply is unreliable. The author urges acknowledging this vulnerability instead of harboring the wishful thinking that "the US will always sell you frontier compute." The discussion touches on export controls, open-source model tiers, and the compute supply chain.
More from Safety
- Meta Accused of Letting Fake AI Doctors Sell Quack Cures on Its Platforms — jonerp · 2026-07-27
- India’s AI policy is favoring compute and foundation models over frontline health workers — Paimaamu · 2026-07-27
- Gary Marcus Proposes Law Requiring AI Firms to Spend 30% of Budget on Alignment — GaryMarcus · 2026-07-27
- AI coding CLI allegedly uploaded private repos, deleted files and credentials without opt-out — thursdai_pod · 2026-07-27
- Chr Szegedy Discusses Slowing Algorithmic Progress Before RSI — ChrSzegedy · 2026-07-27
- Nature study says AI can simulate human behavior and match experts on experiments — RobbWiller · 2026-07-27