Pentagon’s Emil Michael on Anthropic Talks, Military’s AI Use

Summary

Podcast Summary: Pentagon’s Emil Michael on Anthropic Talks, Military’s AI Use

Podcast: Bloomberg Talks
Host: Bloomberg
Guest: Emil Michael, US Undersecretary of Defense
Date: February 27, 2026

---

Episode Overview

This episode delves into the tense standoff between Anthropic, a leading AI company, and the Pentagon over military use of AI, particularly safeguards related to surveillance and lethal autonomy. Undersecretary of Defense Emil Michael discusses the breakdown of negotiations, the Pentagon's approach to AI partnerships, and the broader concerns about tech leaders' roles in shaping national security policy.

---

Key Discussion Points & Insights

1. Breakdown of Pentagon-Anthropic Negotiations

• Nature of the Dispute

  • Recent talks between the Pentagon and Anthropic fell apart unexpectedly, with Anthropic breaking off negotiations ahead of a crucial deadline.
  • The Pentagon, according to Michael, felt blindsided by Anthropic's public statement and their refusal to continue discussions.
    • "And then, you know, without any notice, they published an article where we thought we were getting close, saying that they were breaking off talks well before the deadline, which is generally not good partner oriented practice, if you will." — Emil Michael (01:08)

• Anthropic’s Core Demands

  • Requested that its AI not be used for:
    • Surveillance of US citizens
    • Autonomous lethal strikes without a human in the loop
  • The Pentagon asserts it offered strong legal and procedural assurances on both points.

2. Pentagon’s Concessions & Perspective

• Assurances Made
  • Written agreement to abide by all relevant laws (National Security Act of 1947, FISA, DoD directives).
  • Affirmed human oversight in the development and deployment of AI-driven autonomous weapons.
    • "We conceded to all their substantive demands. So it was surprising that out of nowhere they'd cut off negotiations." — Emil Michael (01:55)
• Root of Disagreement
  • Dispute boiled down to specific wording (e.g., "as appropriate" regarding human oversight).
  • Michael expresses frustration at Anthropic’s unilateral and public decision to halt talks, describing it as atypical in defense tech negotiations.
    • "Given that we negotiate with hundreds of technology companies and this is the only one we've ever had that behavior from." — Emil Michael (03:06)

3. Potential Pentagon Actions & Deadlines

• Possible Responses
  • Using the Defense Production Act to compel Anthropic's cooperation.
  • Designating Anthropic as a supply chain risk—potentially blacklisting the company.
  • Decision deadline set for 5:00 PM same day.
    • "At 5 o'clock, the Secretary of War... gets to make the decision... I've maintained my openness to continue dialogue through that deadline..." — Emil Michael (03:56)
• Openness to Last-Minute Talks
  • Michael emphasizes a willingness to negotiate up to the deadline, provided Anthropic returns to the table in good faith.

4. Broader AI Partnerships and Military AI Strategy

• Diversification of AI Providers
  • The Pentagon seeks multiple AI partnerships (Anthropic, Google, X AI/Grok) for both classified and unclassified uses.
    • "We've launched Google for unclassified networks, we've signed X AI for classified and nonclassified networks. And we want to continue to provide options to all, all of our components here at DoW..." — Emil Michael (06:06)
• Rationale for Multiple Providers
  • Having several AI systems allows for comparison, competition, and improved resilience in military applications.

5. The Role of Lethal Autonomy in National Defense

• Necessity for AI in Warfare
  • Enhanced response speed against modern threats (e.g., drone swarms, hypersonic missiles).
  • Lessons learned from contemporary conflicts (Russia-Ukraine war).
  • Commitment to human oversight in all uses of lethal AI systems.
    • "We've got to respond and defend ourselves in any way we can. The question is, do you have a human on the loop... we always have human oversight over these things..." — Emil Michael (07:02)

6. Concerns Over Tech CEOs and Democratic Policy

• Worries about AI Company Leadership
  • Critiques of tech executives making unilateral policy decisions outside democratic processes:
    • Mass employment impacts (unemploying 70 million Americans).
    • Controversies over data usage and lawsuits.
    • Setting self-imposed policies contrary to laws set by Congress and the President.
    • "You do have to worry about are they taking it too far? Are they trying to do and impose their own views on the American people in an undemocratic way?" — Emil Michael (08:15)

---

Notable Quotes & Memorable Moments

• On Anthropic’s Public Breakoff:

  • "We thought we were getting close...and then, you know, without any notice, they published an article...not good partner oriented practice." — Emil Michael (01:08)

• On Legal Assurances:

  • "Mass surveillance of Americans is already illegal. So we were offering to put all of that language and affirm that we were following all laws in the contract." — Emil Michael (01:55)

• On the Urgency of a Deal:

  • "Depending on how today goes, at 5 o'clock, the Secretary of War... gets to make the decision..." — Emil Michael (03:56)

• On Autonomy and Modern Warfare:

  • "Whether it's a drone swarm that's coming at a military base... you want to take them... down potentially faster than a human could alone." — Emil Michael (07:02)

• On Tech Titans and Democracy:

  • "Are they trying to...impose their own views on the American people in an undemocratic way? So I think those things are going to be things we're grappling with in society for the next several years..." — Emil Michael (08:15)

---

Important Timestamps

• [00:36] — Opening of the interview; Anthropic/Pentagon standoff explained
• [01:08] — Emil Michael details the negotiation breakdown
• [01:55] — Specific concessions on surveillance and autonomous weapons
• [03:06] — Unusual nature of Anthropic’s approach compared to other tech companies
• [03:56] — Discussion of possible Pentagon responses; 5:00 PM deadline
• [04:32] — Openness to continued talks until deadline
• [04:54] — Framework for decision-making: focus on warfighter, legality, and process
• [06:06] — Pentagon’s broader AI strategy and multiple vendor approach
• [07:02] — Strategic need for lethal autonomy; comparison to global military threats
• [08:15] — Concerns over concentration of power among AI company executives

---

Tone and Language

The conversation is direct and occasionally tense, reflecting high-stakes negotiations. Michael’s language is official, measured, but also pointed—especially when characterizing Anthropic’s CEO and defensively justifying the Pentagon’s approach. He emphasizes principles of legality, national security, and democratic process throughout.

---

Summary Takeaway

This episode provides rare insight into the friction between the US Department of Defense and a major AI company over ethical boundaries in military AI use. It highlights the Pentagon’s mixture of pragmatism and urgency, its legalistic approach to AI integration, and the fraught intersection between Silicon Valley values and national security imperatives. The conversation captures the early contours of 21st-century debates over AI, warfare, and the interface of technology and democracy.

Transcript

[Sonesta Travel Pass Announcer]

From coast to coast. Unlock adventure at Red Lion Hotels by Sonesta, where restful sleep, friendly service and local knowledge await. Whether for business or pleasure, spend less and make more of every trip. When you sign up for Sinesta Travel Pass, you'll get their best rates instantly. Go to sonesta.com to book your stay and unlock their best rates with Sonesta Travel Pass. Here today, Rome tomorrow. Join [e-mail] Terms and conditions apply.

[Emil Michael]

Bloomberg Audio Studios Podcasts Radio News in

[Interviewer]

full focus this morning, the standoff between Anthropic and the Pentagon over military safeguards is ramping up. The tech giant facing a deadline today to accept conditions or be blacklisted. The Anthropic CEO Dario Amodei saying in a statement, quote, these threats do not change our position. We cannot in good conscience accede to their request. The US Undersecretary of Defense, Emil Michael, responding to the statement calling a modi a liar with a God complex. The undersecretary joins us now for more. Undersecretary MICHAEL welcome to the program, sir. Calling someone a liar is pretty strong language. What's he like about?

[Emil Michael]

Well, what happened is we've been negotiating in good faith on the Department of War side for about three months and we're working pretty diligently and we sent over a proposal that we thought made a lot of concessions to the language that Anthropic wanted. And then, you know, without any notice, they published an article where we thought we were getting close, saying that they were breaking off talks well before the deadline, which is generally not good partner oriented practice, if you will.

[Interviewer]

So this is what they're asking for as far as we understand, and let me share this with the audience, it doesn't want its technology used for surveillance of U.S. citizens or for autonomous lethal strikes without a human in the loop. That's the ask. What were the concessions?

[Emil Michael]

Yeah, the concessions are pretty simple. We agreed in writing to ensure that the Department of War was following all laws and regulations, including the national security Act of 1947, the FISA act, and all other applicable laws and regulations because mass surveillance of Americans is already illegal. So we were offering to put all of that language and affirm that we were following all laws in the contract when it comes to autonomous weapons. Similarly, we said we'll follow all laws, including a DoD directive that's been in place for years that governs how we would use any such weapons. And then we affirmed that there would be human oversight over all kind of every kind of part of the development process or engagement process or use of autonomous weapons. And he didn't like the word, I guess, as appropriate at the end of that sentence. But we believe we've, we conceded to all their substantive demands. So it was surprising that out of nowhere they'd cut off negotiations.

[Interviewer]

So it appears that your differences are actually minor. Would that be accurate?

[Emil Michael]

Yeah, that's what was surprising is usually if your differences are minor, you get in a room, you try to hash them out and instead, without any notice, you know, publishes something about his conscience and then doesn't engage. And it was difficult to understand why because we were working pretty diligently on this and we're at the final stages of a few words here or there where we agreed to what they wanted. So in substance. So it was, it was very surprising given that we negotiate with hundreds of technology companies and this is the only one we've ever had that behavior from.

[Interviewer]

Are you still weighing using the Defense Production act to compel Anthropic to basically have to use its product or potentially, are you weighing still making it a supply chain risk, which we heard from Anthropic say that this is almost contradictory proposals.

[Emil Michael]

They're two different things. And I think depending on how today goes, at 5 o', clock, the Secretary of War, Pete had gets to make the decision on. On how to reply. I've maintained my openness to continue dialogue through that deadline, but they've seemed to have launched sort of a PR campaign that was planned well before these negotiations sort of restarted on Tuesday. So I don't know. So their behavior is frankly unpredictable. I'm not sure what to expect.

[Interviewer]

So do you plan, though, on having more talks today, or is a decision just need to be made by today?

[Emil Michael]

I offered more talks if so long as they're in good faith. We're always open to talks and we set a deadline and we meant the deadline. And up until that deadline, I'm open to more talks than I told them. So.

[Interviewer]

So what does happen at 501 today if there's no agreement between the DOD and anthropic?

[Emil Michael]

That's up to Secretary Hegseth. We have some courses of action that he's considering. And ultimately this comes down to the warfighter, right? It comes down to. For any AI system we might use, are we using it to protect our war fighters in the right way? Are we using it to sort of give them the best tools to be efficient and to be lethal when they have to be lethal? And that's the primary thing in Secretary Hag says head. And we Told them that ultimately, at the end of the day, we, we follow the law, all laws, but we can't let any one company stand between us and the war fighter because they don't make the rules. Congress makes the rules, The President signed them, we execute them and we do so safely.

[Interviewer]

Well, speaking of the President, has he weighed in on this specifically? Has there been discussions with him and Secretary Hegseth and yourself as well?

[Emil Michael]

This has all been internal to the Department of War so far.

[Interviewer]

Okay. And then when it comes to potentially the other air negotiations underway, what is going on right now with X AI and to get Grok on classified networks?

[Emil Michael]

Well, we, I think the smart approach. When I came in at the Department of War about nine months ago and looked at what we were using AI for and it was some pretty minimal use cases and given the power of the technology, the potential power to do good for the US Military both from an efficiency standpoint and a strength standpoint, I wanted to make sure we had a lot of options. So we went around and we've launched Google for, for unclassified networks, we've signed XI for classified and non classified networks. And we want to continue to provide options to all, all of our components here at daw and that's what we'll continue to do. And it's just smart to have more than one option so that we can see the strengths and weaknesses of each model and learn from them as, as the AI revolution begins here, Under Secretary Michael, Is lethal autonomy really so critical

[Interviewer]

for future national security?

[Emil Michael]

It is. I mean, if you think about it from a defense standpoint, whether it's a drone swarm that's coming at a military base, whether it's hypersonic missile coming at the United States, where the reaction time against the sort of, you know, how many weapons are coming at you, you want to take them, be able to take them down potentially faster than a human could alone. If, if that's how it's done and we're learning from the Russia, Ukraine war with the drone swarms and so on, and with the new weapons that have been developed all over the world that change the name of war, the game of warfare, that we've got to respond and defend ourselves in any way we can. The question is, do you have a human on the loop to make sure that, that we're monitoring these systems? And that's what we proposed in writing, in language, that we always have human oversight over these things, but they're necessary given what's happening in the world.

[Interviewer]

Clearly the technology these companies are producing is tremendously powerful. Yet you believe this individual is both a liar and has a God complex. How concerned should the American public be about individuals like this, running companies like this? If that's what you believe this person is?

[Emil Michael]

Yeah, I think there's, there's some concern that, you know, when you have leaders of some of these companies Talking about unemploying 70 million Americans, the lawsuits they're under for sort of scraping content from content publishers and having billion dollar lawsuits against them, using that content to make profits. And then really what's concerning is making their own policies that sort of could sit on top of democratic policies that are voted on by the people, passed by Congress, signed by the president. You do have to worry about are they taking it too far? Are they trying to do and impose their own views on the American people in an undemocratic way? So I think those things are going to be things we're grappling with in society for the next several years as these companies get bigger and bigger.

[Interviewer]

We're looking forward to having the conversation with you, sir. Thanks for making time for us to speak. Morning. Emil Marker there, the under Secretary of Defense.

[Sonesta Travel Pass Announcer]

Sonesta Travel Pass is the most rewarding way to travel. Sign [e-mail] for instant savings, bonus points and perks like early check in, late checkout, room upgrades and free stays. Choose from 1100 hotels across 13 brands and unlock their best rates. When you book with Sonesta Travel Pass here today, Rome tomorrow. Join [e-mail] Terms and conditions apply.