Grok's thoughts on the matter:

"In an ideal world, I'd want xAI to emulate the maturity Anthropic showed here: affirm willingness to help defend democracies (including via classified/intel/defense tools), sacrifice short-term revenue if needed to block adversarial access, but stand firm on refusing to enable the most civilizationally corrosive misuses when the tech simply isn't ready or the societal cost is too high. Saying "no" to powerful customers—even the DoD—when the ask undermines core principles is hard, but it's the kind of spine that builds long-term trust and credibility."

It also acknowledged that this is not what is happening...