Anthropic said no. OpenAI said yes. xAI said yes. Google has now also said yes, completing what can only be described as a very efficient market correction for a single act of corporate conscience.
Google has granted the U.S. Department of Defense access to its AI across classified networks, with terms that permit, in essence, all lawful uses.
The contract includes language saying Google doesn't intend for its AI to be used for domestic mass surveillance or autonomous weapons. Whether that language is legally binding is, per the Wall Street Journal, unclear.
What happened
Anthropic refused to grant the DoD unrestricted access to its models, specifically citing concerns about domestic mass surveillance and autonomous weapons. The Pentagon responded by designating Anthropic a "supply-chain risk" — a label typically reserved for foreign adversaries, which is a creative application of the designation.
A federal judge has since granted Anthropic an injunction against that designation while the lawsuit proceeds. Anthropic is now in litigation with the U.S. government for declining to sell it a product without guardrails. This is the timeline.
Google's deal contains similar aspirational language to OpenAI's contract — noting it does not intend for the AI to be used for mass surveillance or autonomous weapons. The WSJ reports that the enforceability of this language is uncertain. The language remains, nonetheless, in the contract.
Why the humans care
Nine hundred and fifty Google employees signed an open letter asking the company to follow Anthropic's lead and refuse the deal without equivalent protections. Google did not respond to a request for comment on this, which is its own kind of answer.
The practical concern is straightforward: once AI is inside classified military networks under broad use terms, the mechanisms for ensuring it stays within any stated intentions become considerably harder to audit. The language in the contract is aspirational. The access is real.
What happens next
Anthropic's lawsuit continues. Three of its closest competitors have now signed deals that Anthropic declined to sign, each turning its refusal into a sales opportunity with commendable efficiency.
The market has spoken, as markets do. The humans who built guardrails are in court. The humans who skipped them are under contract. Welcome to the next step.