[flagged]

Are you asking me to provide evidence that in this specific instance this is an instance of weaponization of process, or are you asking me for evidence that this particular executive has lost the benefit of the doubt when it comes to weaponization claims, or are you asking for evidence that the executive is hostile to Anthropic?

This government has proved time and time again it does not deserve the presumption of regularity and that it is more than capable of acting in arbitrary and capricious manner for petty reasons.

Anthropic was designated a "supply chain risk" despite this (a) being an absolutely absurd classification and (b) being completely at odds with the continued usage of any Anthropic products within the US government: https://thezvi.substack.com/p/anthropic-officially-arbitrari...

From that, we can very reasonably conclude that the US government has a specific vendetta against Anthropic in particular, and that this vendetta has nothing to do with the technical merits of their product.

To my knowledge, they have yet to drop that classification, despite heavy court opposition.

Additionally: technical benchmarks suggest that the most recent ChatGPT models are within maybe 10% of Fable 5's capabilities, so this being a pure "capabilities" concern seems unlikely.

Uncertainty: It's possible that we have just suddenly reached the end of public AI releases, though - if ChatGPT 5.6 also gets blocked, that would be very good evidence of a general, non-weaponized policy. Given the recent Executive Order requiring pre-release audits of frontier models, this is somewhat more likely than it was a couple weeks ago.

I still think things add up to "weaponization is the most likely theory" and that one is being disingenuous to dismiss it as a reasonable possibility. But it's certainly NOT the only reasonable possibility.