Article · 2026-07-12 · 6 moments

6 months to live for open models

The most serious test to date of open source AI’s viability is happening right now. ✦ AI generated

GUEST person familiar with the session (via Nathan Lambert)HOST Nathan Lambert

Claim◆

A flat-out ban on frontier open-weight models is not the right policy answer, because if China doesn't also ban them, bad actors can still access banned models, negating any safety benefit.

The author argues a unilateral U.S. ban on frontier open models would fail to improve safety since bad actors could still obtain the models globally, while legitimate U.S. developers and companies would be cut off.

transcript

Nathan Lambert: We need to figure out the right policy for open frontier capabilities, but a flat out ban is likely not the answer. If the models are not banned in China as well, it is VERY easy for a bad actor to still use said banned open weight model, which negates the safety potential.

Fact

At a June 9th White House meeting, a Reflection AI representative argued that open-source models should be exempted from the regulatory framework based on their current capabilities.

A person familiar with a June 9th policy session said a Reflection AI representative pushed for open-source models to be exempted from the framework, noting Reflection has not yet launched a public model.

transcript

person familiar with the session (via Nathan Lambert): A representative from Reflection AI, a U.S.-based open-source model provider, argued that open-source models should have exemptions from the framework based on their capabilities, the person said. Currently, Chinese open-source models such as DeepSeek have a substantial lead over other available open models, and Reflection has not yet launched a public model.

Prediction

A U.S. company like Microsoft or Meta releasing a similarly capable open model soon would shift the narrative away from 'only China builds open models via distillation' and serve as a short-term off-ramp from the current policy death spiral.

The author proposes that a U.S. company releasing a competitive open-weight model would reframe the debate as collaborative rather than a China-vs-U.S. distillation issue, urging Microsoft and Meta to act quickly.

transcript

Nathan Lambert: This is an existential priority for open-source, and the companies who have the business reasons to release open-weight models like Microsoft and Meta (commoditizing their complements) should do it ASAP. I have less faith in Meta, with the new leadership, but they benefit from mass access to AI and should not get in their own way.

provides context · 1

Claim◆

The distillation debate is largely a regulatory capture campaign, since the only proposed solutions massively benefit the organizations pushing for them, particularly Anthropic.

The author argues that Anthropic's anti-Chinese-model distillation campaign has evolved into textbook regulatory capture, since a ban on Chinese open models would directly benefit Anthropic's commercial position.

transcript

Nathan Lambert: 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.

gives example · 1

Claim

The idea that only open-weight models are insecure while APIs are safe is overblown, since APIs have repeatedly been jailbroken and accessed in unintended ways, including Claude Mythos during its private beta.

The author points to repeated jailbreaks and unauthorized access of Anthropic's own Mythos API to argue that the narrative of secure closed APIs versus insecure open weights is overstated.

transcript

Nathan Lambert: I'm not a cybersecurity expert, but the dichotomy that only open-weight models are insecure and APIs are safe has been very overblown in recent years. APIs in concept should be able to be more secure, but that is yet to be demonstrated.

supports · 1

Prediction◆

The most likely incoming action is to ban or indefinitely delay any open-weights model meaningfully above the capability level of GPT 5.5, Claude Opus 4.8, or GLM-5.2, and this will likely happen within the next 6 months.

The author predicts U.S. regulators are likely to ban or indefinitely delay any open-weight model that meaningfully exceeds current frontier capability levels within roughly six months.

transcript

Nathan Lambert: The most likely incoming action is to ban or indefinitely delay any open-weights model meaningfully above the capability level in the range of GPT 5.5, Claude Opus 4.8, or GLM-5.2. With the consistent capability gap, this should be within the next 6 months.

Highlight slides

Likely U.S. Action: Ban or Delay Open-Weight Frontier Models✦ from: The most likely incoming action is to ban or indefinitely delay any open-weights model meaningfully above the capability level of GPT 5.5, Claude Opus 4.8, or GLM-5.2, and this will likely happen within the next 6 months.From Business Concern to Regulatory Capture✦ from: The distillation debate is largely a regulatory capture campaign, since the only proposed solutions massively benefit the organizations pushing for them, particularly Anthropic.Why a Flat-Out Open Model Ban Fails✦ from: A flat-out ban on frontier open-weight models is not the right policy answer, because if China doesn't also ban them, bad actors can still access banned models, negating any safety benefit.Why This Timeline✦ from: The most likely incoming action is to ban or indefinitely delay any open-weights model meaningfully above the capability level of GPT 5.5, Claude Opus 4.8, or GLM-5.2, and this will likely happen within the next 6 months.Who Benefits From the Proposed Fix✦ from: The distillation debate is largely a regulatory capture campaign, since the only proposed solutions massively benefit the organizations pushing for them, particularly Anthropic.The Safety Logic Breaks Down✦ from: A flat-out ban on frontier open-weight models is not the right policy answer, because if China doesn't also ban them, bad actors can still access banned models, negating any safety benefit.

Related episodes

Video · The Cognitive Revolution

Fable's Back, AI Engineer Recap, & SambaNova

→

Video · All-In Podcast

OpenAI vs Anthropic IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts

→

Video · BG2 Pod

The SpaceX IPO, Fable 5, AI Capex Update & Market Check w/ Gavin Baker, Andrew Fox & Clark Tang

→