6 months to live for open models
Interconnects · Nathan Lambert · 2026-07-12
Nathan Lambert warns that open-weight AI models face potential U.S. regulatory bans within six months, driven by a capability threshold tied to frontier models and an Anthropic-led campaign against Chinese AI distillation that Lambert characterizes as regulatory capture.
Appears in
Extraction
Topics: open-source-aiai-regulationai-policydistillationchinese-ai-models
Claims
- White House discussions are underway on an executive order that could ban or indefinitely delay open-weight models above the capability level of current frontier closed models, likely within six months.
- Anthropic's campaign against Chinese model distillation constitutes regulatory capture because Anthropic would gain direct economic benefit if Chinese open-source competitors were banned.
- Model APIs are not meaningfully more secure than open weights, as demonstrated by unauthorized Discord access to Anthropic's Mythos model even during private beta.
- A unilateral U.S. ban on open-weight models would be ineffective because bad actors could still access those models through Chinese or other jurisdictions.
- The most viable short-term intervention is for U.S. companies like Meta or Microsoft to release competitive open-weight frontier models, shifting the narrative away from the framing that only Chinese labs produce capable open models.
Key quotes
A ban of any form here would be a big mistake for the long-term trajectory of AI.
This campaign may have started through a genuine business concern, but it has progressed to be the definition of regulatory capture, as Anthropic would gain substantial economic security in its products if the Chinese model makers they accused were banned.
Open models increase safety by broad access and understanding, not kneecapping the positive actors only.