Alvaro Felipe Chávez

Between the lines, Anthropic is saying “I’m okay with open weights as long as the model does not present high performance.” “Release what you want as long as not a direct alternative to proprietary models.”

DARIO DAMAGE CONTROL WARNING! I BROUGHT THE RECEIPTS TO SHOW THE GUARDRAILS OF TRUTH. Up next is a textbook damage control operation dressed up as consistency. Dario Amodei’s July 27, 2026 post (“Our position on open-weights models”) claims Anthropic “has never advocated for a ban on open-weights models” and that open-weights models without dangerous capabilities “are a public good.” Anyone who has tracked his actual record knows this is a carefully worded retreat under pressure after Anthropic refused to sign the Nvidia-led open-weights letter and faced accusations of protectionism. The 2023 Senate receipts are unambiguous On July 25, 2023, before the Senate Judiciary Subcommittee on Privacy, Technology, and the Law, Amodei said the following: “I think in most scientific fields, open source is a good thing… and I think even within AI there’s room for models on the smaller and medium side… And I think to be fair, even up to the level of open source models that have been released so far, the risks are relatively limited… But I’m very concerned about where things are going… I think the path that things are going, in terms of the scaling of open source models—I think it’s going down a very dangerous path. And if that path continues, I think we could get to a very dangerous place.” He continued: “When you control a model and you’re deploying it, you have the ability to moderate usage. It might be misused at one point, but then you can alter the model, revoke a user’s access, or change what the model is willing to do. When a model is released in an uncontrolled manner, there’s no ability to do that. It’s entirely out of your hands.” And on the distinction that matters: “Open source normally refers to smaller developers who are iterating quickly, and I think that’s a good thing. But what we’re talking about here is something a little different: the uncontrolled release of much larger models by big entities that pay tens or even hundreds of millions of dollars to train them. I think we should treat those in a somewhat different category, with different obligations.” His written testimony the same day warned that bad actors could “repurpose open-source models if powerful enough open-source models become available.” That is not neutral language. That is a clear, public warning to Congress that the trajectory of open-sourcing increasingly capable models is dangerous and irreversible once the weights are out. He was not talking about toy models. He was talking about the scaling path toward frontier systems. The 2026 post does not square with that record In the new post Amodei writes: “Anyone who has read my past writing should know that I don’t regard such bans as a useful measure, but let me state it clearly so that there is no doubt: Anthropic has never advocated for a ban on open-weights models.” and “Open-weights models that don’t have dangerous capabilities are a public good…” This is a motte-and-bailey. The 2023 testimony was not a narrow objection to “bans on Chinese models used by U.S. businesses.” It was a broad claim that the path of scaling open models itself was “very dangerous” because release is permanent and control is lost. Calling non-dangerous open-weights a “public good” while still insisting powerful open-weights carry higher cyber/bio/alignment risk (exactly the same irreversibility argument he made in 2023) does not erase the earlier framing. It reframes it once the political cost of the earlier framing became high. He further says a ban on use by U.S. businesses “would not address” his concerns and “has never been my goal.” 1 of 2

false equivalence. anthropic wants to ban open source models

Attackers have frontier AI. Defenders need a frontier AI ecosystem—the best open and closed models, force-multiplied by a global community. During the Hugging Face incident, closed AI blocked essential forensics. An open-weight frontier model helped contain the intrusion. That’s why we created the Open Secure AI Alliance.

If that OAI model that hacked HF were open weight, anyone with a few chips could run it safeguard free! If OAI fixes the alignment, anyone with a few more chips could tune the alignment out! Yes, anyone can identify vulnerabilities! And if a16z commits to patching every hospital

To crystallize my objections to the open weight letter: “Open weight models, on the other hand, allow a broad community of researchers and developers to examine their behavior, identify vulnerabilities, develop safeguards, and improve them over time.” This is just not true!

OpenAI has signed now. What a fascinating day of game theory.

julian is clearly right in the spirit of his tweet: nvidia and microsoft aren’t unbiased in their support of oss, it’s majorly motivated by competitive incentives. good take and glad he said it