There you have it! The Pentagon calls Anthropic a "supply-chain risk" and locks them out of defen…
June 30, 2026 · 0 likes · 0 comments
China Threat AI Defense Cybersecurity
There you have it! The Pentagon calls Anthropic a "supply-chain risk" and locks them out of defense contracting entirely.
California just signed them up for every state agency. At half price. With training and support included.
Same company. Same week. Two governments. Opposite verdicts.
I have spent two years documenting this exact story, post after post. Washington pulled Anthropic's frontier models offline over a weekend. Slapped them with the "supply-chain risk" label reserved for Huawei and Kaspersky. Refused their contract because Anthropic wanted written carve-outs against surveilling Americans and fielding autonomous weapons without a human in the loop. The DoD signed with OpenAI instead.
Then the same federal government barred Anthropic from working with any other Pentagon contractor.
And while all of that was happening, Gavin Newsom handed Anthropic the keys to the most populous state in the country.
Here is the part that should stop you cold.
California's own CIO, Chris Given, told Politico the federal "supply-chain risk" designation "just didn't come up" during negotiations.
Didn't come up.
The federal government formally flags a vendor as a national-security risk. The largest state in the union puts that vendor inside every agency and every local government. And the people doing the buying say the flag never entered the conversation.
So which is the truth? Either the Pentagon's designation means something — in which case California just walked its government data into a vendor Washington won't touch — or it means nothing, in which case the "supply-chain risk" label was never about security in the first place.
It cannot be both. A real risk doesn't stop at a state line.
I am not even arguing California's deal is wrong. Document drafting and analysis is a narrow use case. The vendor lock-in concern is real — half price today is full price in year three — but that is a normal procurement debate.
The scandal is the contradiction. We have built a system where the same company is a national-security threat in one building and a discount partner in the building across the street, and nobody can tell you why. No published standard. No legal test a state can read and apply. Just a label that means everything in D.C. and nothing in Sacramento.
This is exactly the chaos that happens when access to American AI gets decided by phone calls instead of rules. I have spent my career inside this world, and I know what it costs: while we argue with ourselves about which of our own companies to trust, China ships open-source models nobody can revoke, label, or switch off.
I only found this story because I read it on UnbiasedHeadlines.com — the site my AI agents built to cut through exactly this kind of spin. Facts, sources, both sides, no thumb on the scale. The original briefing is there if you want every detail.
What are your thoughts?
California just signed them up for every state agency. At half price. With training and support included.
Same company. Same week. Two governments. Opposite verdicts.
I have spent two years documenting this exact story, post after post. Washington pulled Anthropic's frontier models offline over a weekend. Slapped them with the "supply-chain risk" label reserved for Huawei and Kaspersky. Refused their contract because Anthropic wanted written carve-outs against surveilling Americans and fielding autonomous weapons without a human in the loop. The DoD signed with OpenAI instead.
Then the same federal government barred Anthropic from working with any other Pentagon contractor.
And while all of that was happening, Gavin Newsom handed Anthropic the keys to the most populous state in the country.
Here is the part that should stop you cold.
California's own CIO, Chris Given, told Politico the federal "supply-chain risk" designation "just didn't come up" during negotiations.
Didn't come up.
The federal government formally flags a vendor as a national-security risk. The largest state in the union puts that vendor inside every agency and every local government. And the people doing the buying say the flag never entered the conversation.
So which is the truth? Either the Pentagon's designation means something — in which case California just walked its government data into a vendor Washington won't touch — or it means nothing, in which case the "supply-chain risk" label was never about security in the first place.
It cannot be both. A real risk doesn't stop at a state line.
I am not even arguing California's deal is wrong. Document drafting and analysis is a narrow use case. The vendor lock-in concern is real — half price today is full price in year three — but that is a normal procurement debate.
The scandal is the contradiction. We have built a system where the same company is a national-security threat in one building and a discount partner in the building across the street, and nobody can tell you why. No published standard. No legal test a state can read and apply. Just a label that means everything in D.C. and nothing in Sacramento.
This is exactly the chaos that happens when access to American AI gets decided by phone calls instead of rules. I have spent my career inside this world, and I know what it costs: while we argue with ourselves about which of our own companies to trust, China ships open-source models nobody can revoke, label, or switch off.
I only found this story because I read it on UnbiasedHeadlines.com — the site my AI agents built to cut through exactly this kind of spin. Facts, sources, both sides, no thumb on the scale. The original briefing is there if you want every detail.
What are your thoughts?