| then they went back and said no, you need to remove those safeguards to which Anthropic is (rightly so) saying no.

So one thing to call out here is that the assumption that DoW is working on specifically these use cases is not bullet proof. They simply may not want to share with anthropic exactly what they are working on for natsec issues. /we can't tell you/ could violate the terms.

It is also dumb that DoW accepted these terms in the first place.