Interview with Pentagon AI head Emil Michael on his view that Anthropic leaked negotiations to the press to win anti-Trump users, dealing with Amodei, and more
what if this software went down? Some guardrail kicked up? Some refusal happened for the next fight like this one and we left our people at risk?” “I went to @SecWar @PeteHegseth and told him what happened.” [video]@gurumedasani:@AnthropicAI should take this guy to court for declaring a major American AI lab a supply chain risk when it isn't. They have clearly been using Anthropic models successfully in Department of Defense missions and now want to punish them. This approach will fail as th
The disagreement is now also a contest over whether Anthropic’s deployment limits make it dependable for defense work. Amodei had publicly framed the company’s concerns as cases where AI capability may move ahead of law and American values.
First-order effects
Emil Michael’s warning makes Anthropic’s guardrails and reliability a live procurement issue for the Department of Defense, while Anthropic faces a public challenge to its suitability for sensitive missions.
The allegation that Anthropic leaked negotiations intensifies the dispute beyond technical policy, putting the company’s motives and its relationship with Pentagon leadership under sharper scrutiny.
Second-order effects
Defense buyers will have greater incentive to test model availability, refusal behavior, and fallback options before assigning frontier models to operational workflows.
Other AI vendors can position themselves as more deployable for national-security use, while labs seeking Pentagon business face pressure to state their boundaries more explicitly.
Third-order effects
If this pattern persists, access to government AI work will increasingly hinge on whether a lab is viewed as state-compatible—not only on model capability or safety claims.
The conflict could harden a split between providers that accept broad defense-use terms and those that preserve tighter deployment constraints, though the eventual boundary will depend on procurement terms and policy decisions.
The trend: Frontier AI is becoming strategic infrastructure whose access, safeguards, and operational reliability are negotiated as matters of state power.
Under Secretary of War for Research and Engineering Emil Michael on being the CTO of the Department of War, applying lessons from Silicon Valley at the Pentagon, and his “holy cow” moment with AI vendors. 00:00 Silicon Valley to DC 02:03 Why the DoW cannot operate at “peacetime […
Gets flustered by bureaucracy, high ego, blames the previous guy, blows up existing supplier (that seemed to be adding value?) b/c of weird hypotheticals, didn't forsee (or care) about public spat. All so avoidable - why are we here??
.@USWREMichael says the Maduro raid was the trigger point for the DoW's conflict with Anthropic: “Palantir's the prime contractor. [Anthropic] is the sub. One of [Anthropic's] execs called Palantir and asked, ‘Was our software used in that raid?’” ... “It raised enough alarm w…
@AnthropicAI should take this guy to court for declaring a major American AI lab a supply chain risk when it isn't. They have clearly been using Anthropic models successfully in Department of Defense missions and now want to punish them. This approach will fail as they will
This is illuminating re: DoW's thinking, but it doesn't remotely justify the decision to declare Anthropic a supply chain risk. 3252 defines “supply chain risk” as follows: “The term ‘supply chain risk’ means the risk that an adversary may sabotage, maliciously introduce
This honestly sounds like @USWREMichael is targeting Anthropic and may be corrupt. What are his ties to the other labs? Who is he colluding with? None of his actions make sense here. If you don't like their policies, just cancel the contract. Simple as that.
The team at @PirateWires have an enlightening interview with @USWREMichael. He explains how the @DeptofWar made the important decision to confront Anthropic over their onerous terms and services that would significantly impair our nation's military readiness. [image]
every single piece of evidence and reporting about the Anthropic/DoW spat has revealed that they were an absolute liability and had no place in our wartime military infrastructure*
Hmmm... @andrewrsorkin has reported on-air for the last few days that OpenAI has the very same problematic terms described, that caused the DOW to part with Anthropic. He says these ToS are really a fig leaf for Anthropic not being MAGA enough in its politics. Which is it?
pirate wires interviewed the DOW's AI chief yesterday. new details here on the negotiation with anthropic, including more context on the SCR designations, and color on a massive culture clash. most salient point, perhaps: emil michael says a deal is still possible.
EXCLUSIVE: Department of War AI Chief On How The Anthropic Deal Collapsed When Emil Michael (@USWREMichael) took over the Department of War's AI portfolio last August, he discovered the Biden admin had been “asleep at the wheel” when it came to top military contracts. “I was [ima…
Inside the Culture Clash That Tore Apart the Pentagon's Anthropic Deal // The craziest thing about this deal (to me) is the idea that there are _any_ terms of use or operational parameters in a deal with DoD. My experience over a very long time is you kill yourself to get a
This clarifies why the DoD is pursuing the supply chain risk designation against Anthropic. The tensions started with the Maduro raid. The DoD was using Palantir as a service provider during the raid. Palantir was using Claude. Anthropic contacted Palantir and started asking
ok, I was unsure & leaning this way so far, but this really cements that the DoW did the right thing re: Anthropic. Sorry guys, love your product but you simply can not act like all of your customers are in the same league. Hopefully you can swallow your pride & move forward.
Nothing in here to justify the SCR decision, though nice to hear the bridge isn't totally burned. One hopes cooler heads will prevail and the designation will be withdrawn sooner rather than later.
i'm fckin exhausted from all the anthropic drama tbh but it keeps getting more and unhinged, timeline of events: - palantir revealed they used claude to capture president maduro - anthropic didn't like that. raises concerns to palantir. - palantir tells pentagon who panics and
Whatever you think of Dario's decision, it's super impressive how employees have closed ranks behind him During heated controversies, companies often have people publicly dissenting, resigning in protest, etc Anthropic's united front is a sign of internal consistency and trust
Nathan is being polite here. Either .@emilmichael, a senior member of the Department of War, is wildly mistaken about major AI policy, or he chose to say something that is wildly untrue.