Devil's advocate, but Anthropic talks a lot about importance of alignment, so.. How can military ensure that model is NOT refusing to work just to be malicious?

Anthropic doesn't provide model w/o guardrails, suppliers have to use public version, newer model releases surely know about Anthropic standoff with the military.. how to prevent model from realizing that it's doing something for the military (supplier) and sandbagging and/or subtly sabotaging the results?

=====

DOD recommends sandbagging as defense mechanism against destination "attacks":

https://media.defense.gov/2026/Sep/08/2003992823/-1/-1/1/CSA...

Page 13

  When suspecting a malicious distillation campaign, consider varying changes to responses across requests to complicate response quality evaluations, such that the subtle changes avoid triggering obvious alerts. Reducing reasoning depth, presenting correct information with different reasoning, or stylistic inconsistencies may evade detection while reducing training usefulness.

  Avoid informing China-based AI company users suspected of distillation campaigns of a switch to a downgraded model. Informing malicious distillers would enable them to improve their defense evasions and indicate when to roll back training.

they issued the supply chain risk to get it out of the supply chain.

they don't want suppliers to use it because it is a supply chain risk. specifically, the risk is that it is unreliable.

the designation stops suppliers from using it.

it stops it from being in the supply chain.

How is this a debate? Anthropic's AI was used in military/defense operations. When Anthropic discovered this and disclosed it publicly, they attempted to revoke access and went after the government demanding it not get used in such a way. This would be a supply chain risk to any government.

If Lockheed Martin disabled weapons systems when it discovered the Air Force using its fighters over Iranian airspace based on Lockheed's ideological views, I'm pretty sure they would be labeled a supply chain risk at minimum. What is the issue here?