On Thursday, Anthropic CEO Dario Amodei reiterated his refusal to allow Claude to be used for mass surveillance of U.S. citizens or to guide fully autonomous weapons, rejecting Pentagon requests to make unfettered use of the model.

Claude is one of just two large generative-AI models that the Pentagon has made available on classified networks, and it is the only one that belongs to the cutting-edge group of frontier models. The Defense Department isn’t saying just how it uses such models. But Emil Michael, defense undersecretary for research and engineering, has suggested that their uses include intelligence (“to synthesize a lot more intelligence using a machine than a human analyst”) and warfighting (“How do you predict what might happen in the conflict, what things you might need in the conflict?”).

Earlier on Thursday, Pentagon spokesperson Sean Parnell said that DOD only seeks the ability to “use Anthropic's model for all lawful purposes,” adding that the idea that the Pentagon wants fully autonomous weapons or mass surveillance is a false narrative “peddled by leftists in the media.” 

But Amodei said those are the only two limits he insists on. 

In “a narrow set of cases, we believe AI can undermine, rather than defend, democratic values. Some uses are also simply outside the bounds of what today’s technology can safely and reliably do,” he said in his statement.

Pentagon officials have threatened various reprisals should Anthropic insist on its limits, including invoking the Defense Production Act to use the company’s product without the company’s permission. 

On Wednesday, a defense official told Defense One, “The Secretary will not hesitate to invoke the DPA if an agreement cannot be reached.”

Parnell’s post on Thursday made no mention of the DPA. The company, he said, has “until 5:01 PM ET on Friday to decide. Otherwise, we will terminate our partnership with Anthropic and deem them a supply chain risk for DOW.”

In his statement, Amodei responded quizzically. “They have threatened to remove us from their systems if we maintain these safeguards; they have also threatened to designate us a ‘supply chain risk’—a label reserved for US adversaries, never before applied to an American company—and to invoke the Defense Production Act to force the safeguards’ removal. These latter two threats are inherently contradictory: one labels us a security risk; the other labels Claude as essential to national security.”

Easier said than done

If the Pentagon does designate the San Francisco-based AI startup as a supply-chain risk, it would touch off a lengthy and likely expensive series of protective measures, the people familiar said. 

Operators would have to reconfigure data inputs that they are feeding into models, re-examine how to share data in real-time with the intelligence community which also uses Claude widely, and re-validate that replacement models were functioning as the military expected it to, they said.

In July, Anthropic received a $200 million contract to provide its frontier-model tools to the Pentagon, as did the other three U.S. makers of such products: OpenAI, Google, and xAI. 

Department leaders have urged their people to use the new tools, though they have declined to say how publicly. And even the Pentagon doesn’t really know; it is reportedly asking various commands to describe how much they use Anthropic. (Michael, however, has described U.S. INDOPACOM as “probably one of the premier users.”)

So why is Claude the only one deployed on classified networks? One key reason, according to a defense official: Anthropic’s tools were the easiest to deploy on cloud networks powered by AWS, which contributes the largest chunk of the Pentagon’s Joint Warfighting Cloud Capability.

The two companies are especially close. AWS is the leading cloud-service provider to Anthropic, which trains its models using Amazon’s proprietary Trainium chips.

By contrast, Google runs Gemini on its own cloud and trains it on TPU v5p chips. xAI is partnered with Oracle and does most of its Grok training on NVIDIA H100 GPUs. OpenAI has a “primary” relationship with Microsoft Azure, though it recently announced a “strategic training” partnership with AWS.

None of these relationships are static. Anthropic trained its first models on NVIDIA chips. But as demand grew, the various frontier AI companies inked long-term strategic contracts that mean migrating from one environment to another would undo months of work. 

The individuals said it could be twelve months or longer to replace the capability. However, a Defense Department official said that he expected additional frontier AI models to be widely available on the Pentagon’s GenAi.mil interface before summer. 

AWS did not respond to requests for comment.

Breaking up for the wrong reasons

Michael has said that his objection to Anthropic’s stance is that it creates unpredictability. What if, he said last week, operators were using Claude during a mission, and  “then the model itself learns what you're trying to do… and it stops working. That’s a risk I cannot take.”

But Anthropic executives counter that they must draw lines precisely because of AI’s unpredictability. They say there’s no way to guarantee that their models can perform safely in scenarios that involve lethal autonomy—at least not without meaningful human supervision—and they don’t believe the model is safe in situations that might involve AI for mass surveillance, according to sources familiar with the discussions.

And they agree with Michael’s contention that some of the Pentagon’s frontier models might perform better at various tasks than others.

The sources also said the conversations between the Pentagon and the company had been proceeding along more or less normal lines. Anthropic, they say, had been willing to make various accommodations. But the tone changed after the discussions became public.

On Tuesday, the company released a new version of its safety policies, which many saw as an abandonment of its core safety promise.

In the blog post announcing the change, the company said that it would be moving toward “nonbinding but publicly declared targets” for safety. “Rather than being hard commitments, these are public goals that we will openly grade our progress towards.”

Lawmakers are dipping a toe into the debate. Sen. Mark Warner, D-Va., called the fight “another indication that the Department of Defense seeks to completely ignore AI governance–something the Administration’s own Office of Management and Budget and Office of Science and Technology Policy have described as fundamental enablers of effective AI usage,” in a statement. He called the episode further evidence of “the need for Congress to enact strong, binding AI governance mechanisms for national security contexts.”

The Pentagon has in the past placed policy limits on the use of autonomous weapons, but Congress has passed no legislative limits.

“I appreciate the concern, and I would point to past precedent in a lot of cases. For example, during the COVID pandemic, we did have to utilize National Guard to assist with all the logistics around these things,” Mark Ditlevson, nominated to be the assistant secretary of defense for homeland defense and Americas security affairs, told the Senate Armed Services Committee during his confirmation hearing. If confirmed, “I commit to following or making sure that we analyze this, provide the best recommendation up to the secretary. We’d talk to the office of general counsel to make sure everything that we're doing is legal and lawful. Before that advice gets to the secretary, I can make that commitment to you.”

The National Guard often provides logistical support for elections, including cybersecurity functions. But that’s different from patrolling polling locations, senators argued, pushing Ditlevson for a yes or no answer on whether there are credible threats to the elections that would warrant calling in the National Guard. 

When asked about National Guard troop presence at election polling locations, Ditlevson deferred to the Pentagon's general counsel on the legality of such deployments, acknowledging U.S. law that bars National Guard from election locations unless there are “armed enemies of the United States” present.

Ditlevson, who has been acting in the role since last year, agreed the statute is an “extremely high bar to meet” that requires “robust analysis,” but stopped short of refusing to recommend, as a matter of policy, sending troops to the polls. 

Senators—particularly Democrats—took umbrage with Ditlevson’s opaque answers.

“You are nominated for this position. I'm just asking, do you think it would be appropriate to station troops next to polling stations? Simple yes or no,” said Sen. Elizabeth Warren, D-Massachusetts.

Ditlevson called the question “speculative,” declining to discuss “what threat levels may exist during an election cycle.”

Warren retorted: “I have to say, if you're not willing, just to say, ‘No, it is not appropriate,’ then I have real concerns about you in this job.”

Sen. Tim Sheehy, R-Montana, brought up violence he saw around the 2010 elections in Iraq when he was a Navy SEAL, saying, “the point was to suppress voter turnout, because they didn't want a democratically elected government.”

He asked if it would be appropriate to use the National Guard and other resources in the event of a “direct terrorist threat against a polling station.” 

Ditlevson said yes. 

The debate comes alongside reports the White House is considering an executive order that seeks to codify President Donald Trump’s previous calls to “nationalize” elections. Trump’s supporters then called for immigration agents to stake out the polls during the upcoming midterm elections. A Homeland Security official has since promised that ICE agents, who have violated multiple court orders in Minnesota, won’t be at the polls during the midterms this year. 

In August, Trump deployed the National Guard in Washington, D.C. to fight crime, followed by deployments to several other cities, facing myriad legal challenges. Those deployments, which have been winding down, could cost taxpayers more than $1 billion. 

A recent Associated Press-NORC Center for Public Affairs Research poll found that only 27% of Americans have “a great deal or quite a bit” trust in Trump’s judgement in using military force. Along party lines, 60 percent of Republicans had that same level of confidence compared to 5 percent of Democrats and 14 percent of independent voters. 

Sen. Tammy Duckworth, D-Illinois, closing out the hearing in a fiery exchange, asked several questions: how far from polling stations would National Guard troops be posted for logistics support, would they be in uniform, would they wear Kevlar or would they carry a sidearm. 

“Are you saying that there are potentially other reasons to send them to polling stations, beyond logistics, and that they may actually be fully up-armored, fully combat equipped, depending on what you decide is the situation on the ground?” she said. 

Ditlevson called the question “speculative,” but said “questions for the specific distance and what equipment people would wear on any deployment: We always look at what the threat scenario may be and tailor that for whatever they may have to deal with. And so, as you pointed out, for logistics, it'd be a very different footprint than some other type of mission.” 

Those responses still left senators with questions. 

“I am concerned. I am concerned that you can't make basic reassurances to this committee as to how you would advise on the calling of our National Guard to be there at polling places at an election on United States soil,” Duckworth said. 

“You cannot intimidate Americans when they go out to vote. It is against the law. It is against code. I just read that code to you. I am at a loss,” she said. “The American people would not stand for this. I will not stand for it, and my colleagues in Congress should not stand for this administration's attempts to use our nation's military to intimidate Americans from exercising their fundamental right to cast a vote in elections.”

“Caine hand-delivered memos to [Defense Secretary Pete] Hegseth and the Pentagon’s policy chief Elbridge Colby outlining his disagreements over the new National Defense Strategy that Colby’s office had drafted,” a source told CNN. 

Caine’s reported reservations over the NDS are surfacing amid parallel reports about his advice regarding possible U.S. military strikes against Iran. Axios, the Wall Street Journal and the New York Times recently relayed reports of Caine’s concerns that a new war could be more prolonged and damaging than some Iran hawks in Washington are willing to publicly admit. 

Despite the president’s claims that last June’s strikes inside Iran “obliterated” Tehran’s nuclear program, the president’s envoy Steve Witkoff claimed on Saturday that Iran is now “probably a week away from having industrial-grade bombmaking material.” On Wednesday, CNN’s Kaitlin Collins asked Vice President JD Vance, “Can you explain to the American people why the United States would need to strike Iran to stop them from getting a nuclear weapon if the U.S ‘obliterated. their enrichment program last summer?” 

Vance declined to answer, and replied instead, “I'm not going to make any news on Iran today.” Meanwhile, the administration’s third round of informal U.S.-Iran talks continue today in Geneva, led by Witkoff and Trump’s son-in-law Jared Kushner. Axios has the latest, which at a reported midpoint is inconclusive so far, here. 

A 65-year-old former Air Force officer was arrested this month after allegedly spending more than two years training military pilots in China, according to the Department of Justice, which announced the charges Wednesday. His name is Gerald Brown, and he left active duty in 1996 as a major after 24 years of service. He was arrested Wednesday in Jeffersonville, Indiana, WDRB reports from Louisville, Ky.

While in the Air Force, he served as a fighter pilot instructor and simulator instructor on a variety of military aircraft, including the F-4 Phantom II, F-15, F-16, and A-10. He later worked for two U.S. defense contractors training U.S. military pilots on the A-10 and the F-35. But, allegedly, in August 2023, he began correspondence for the job that would ultimately lead to his arrest. He flew to China four months later, where he began training pilots in the Chinese military. He returned to the U.S. earlier this month, and was arrested shortly afterward.

If his case sounds familiar, “​​​​The charges against Brown follow similar charges filed against former U.S. Marine Corps pilot Daniel Edmund Duggan in the District of Columbia in September 2017,” the Justice Department said Wednesday. Duggan was later charged with violating the Arms Export Control Act and conspiring to engage in money laundering. 

Related: 

• “How China is masking drone flights in potential Taiwan rehearsal,” via Reuters, reporting Thursday from Hong Kong; 
• And “Taiwan should create drone swarm ‘asymmetric hellscape’ to blunt Chinese invasion: Report,” Breaking Defense reported Thursday off new analysis from the Center for a New American Security. 

The Air Force’s top mobility leader is pushing for decades-old air transports and tankers to be replaced sooner than the late 2030s, as currently planned, Defense One’s Thomas Novelly reported Wednesday from the Air and Space Forces Association's Warfare Symposium in Colorado. 

Background: A solicitation memo late last year detailed the service’s plans to keep the C-5 and C-17 flying until 2045 and 2075, respectively. A next-generation airlift is not expected to be fielded until at least 2038. “We must pay attention to that strategic capability,” Lt. Gen. Reba Sonkiss, the interim head of Air Mobility Command, told reporters on Tuesday. We're, again, woefully behind on the modernization front for our strategic air forces.”  

A particular platform of concern: KC-135s. “I cannot have a 90-year-old tanker refueling a B-21, and if you do the math, as we reach the end of programs for things, that’s the reality,” she said. Meanwhile, she praised the C-5 as a “critical tool” but recognized its longstanding problems. The aircraft, first fielded in the 1970s, achieved only a 48 percent mission-capable rate as of 2024. “It is an old airplane,” Sonkiss said. “We have to get after what next looks like, and we can't wait until we're shoveling it into the boneyard before we get to that discussion.” Read more, here. 

Technical update: U.S. defense startup Anduril flew its YFQ-44A collaborative combat aircraft with Shield AI's Hivemind autonomy software as well as its own, the company said Wednesday, marking another step in the Air Force's ongoing drone wingman competition. “The aircraft took off and autonomously approached a designated point where Shield AI’s mission autonomy software stack, Hivemind, was activated to complete a series of test cards,” Anduril said in a news release. “Following completion of Hivemind tests, Anduril was able to seamlessly switch to Anduril’s Lattice for Mission Autonomy stack to complete the same test points, before returning safely to land.”

Anduril announced the development on the last day of the ASFAW symposium. The startup’s YFQ-44A also began integrated weapons testing earlier this month, Novelly reported Monday.

Today on Capitol Hill, Trump’s top Pentagon official responsible for planning National Guard deployments to multiple U.S. cities is testifying before the Senate Armed Services Committee. That’s Mark Ditlevson, who has been nominated as the Pentagon’s next Assistant Defense Secretary for Homeland Defense and Americas Security Affairs. 

But the full scope of his responsibility in that post is much larger. It also includes all defense and security policy for Canada, Mexico, Central America, South America, and the Caribbean. Ranking member Jack Reed of Rhode Island reminded Ditlevson in his opening remarks that “Chairman [Roger] Wicker and I have repeatedly directed the Department to submit basic information—required by law—to the Committee regarding the ongoing boat strike campaign in the Caribbean and Eastern Pacific.” 

“To date, the Department has failed to submit to Congress the Execute Orders and videos related to these operations, and it has refused to make public the legal justifications and intelligence underpinning the strikes,” Reed said. “Given that you were the Acting Secretary of Defense for Homeland Defense and Americas Security Affairs during the build-up of U.S. military forces in the Caribbean, and you continued to hold that position after the Department of Defense launched its campaign of military strikes against alleged drug boats, I expect your commitment today to submit these legally required documents to Congress.”

Former Texas state Sen. Brian Birdwell is joining Ditlevson before SASC. Birdwell is a 20-year Army veteran who participated in the Gulf War and was later wounded during the 9/11 attack at the Pentagon. He’s been nominated as the next assistant secretary for sustainment. Catch a livestream of their testimony, here. 

Additional reading: 

• “U.S. must overhaul military readiness and tech metrics, report urges,” Axios reported Wednesday; 
• “New hypersonic missiles unveiled for fighter planes, ground systems,” Military Times reported Wednesday after a news release from Colorado-based Ursa Major; 
• And “Joint Staff director returning to Navy after only months on job,” Breaking Defense reported Wednesday. 

Welcome to this Thursday edition of The D Brief, a newsletter focused on developments affecting the future of U.S. national security, brought to you by Ben Watson with Thomas Novelly. It’s more important than ever to stay informed, so we’d like to take a moment to thank you for reading. Share your tips and feedback here. And if you’re not already subscribed, you can do that here. On this day in 1945, American forces seized the Philippine island of Corregidor from the Japanese.

Latin American developments

The Cuban coast guard killed four people in a speedboat who had allegedly opened fire upon them Wednesday morning, the Cuban Embassy to the U.S. said in a statement on Twitter. 

The incident occurred roughly one nautical mile from the Cuban shore at Cayo Falcones, which is about 260 miles southeast of Key West. 

Narrative of the encounter: “When a surface unit of the Border Guard Troops of the Ministry of the Interior, carrying five service members, approached the vessel for identification, the crew of the violating speedboat opened fire on the Cuban personnel, resulting in the injury of the commander of the Cuban vessel,” the embassy said in its statement. “Four aggressors on the foreign vessel were killed and six were injured. The injured individuals were evacuated and received medical assistance.”

While the boat was registered in Florida, Cuba was careful not to claim its occupants were American. And later Wednesday, the Cuban government said all 10 in the speedboat were “armed Cubans living in the U.S. who were trying to infiltrate the island and unleash terrorism,” the Associated Press reports. 

Most of the 10 “have a known history of criminal and violent activity,” Cuban officials said. At least one of the four killed, Michel Ortega Casanova, “was a truck driver and an American citizen who lived for more than 20 years in the U.S.,” his brother told AP. He said his brother had become “obsessive and diabolical” in pursuit of freedom for Cuba. 

State Department reax: “We’re going to have our own information on this,” Secretary of State Marco Rubio told reporters Wednesday on the sidelines of a regional conference at a Marriott resort in the Caribbean. “We’re not going to base our conclusions on what they’ve told us,” he said of the account from Cuban officials. He also declined “to speculate about whose boat it was, what they were doing, why they were there, what actually happened.”

“Suffice it to say it is highly unusual to see shootouts on open sea like that,” Rubio said, and noted, “we do have constant contact with them at the Coast Guard level where they notify the U.S. Coast Guard on a variety of things, including migrants and so forth.” 

The U.S. has expanded Venezuela oil exports to Cuban businesses—just not to the Cuban government, the Treasury Department announced Wednesday. The change in licensing policy applies to Cuba’s “private sector (e.g., exports for commercial and humanitarian use in Cuba)” but does not extend to “the Cuban military, intelligence services, or other government institutions.” 

According to Rubio, “These would be sales to a very small private sector that exists in Cuba, and that’s always been legal,” he said Wednesday. “This would just expand to the numbers that could do it.”  However, he warned, “If we catch the private sector there playing games and diverting it to the regime or to the military company, if we find that they’re moving that stuff around in ways that violate the spirit and the scope of these permissions, those licenses will be canceled.” 

By the way: Back stateside, abducted Venezuelan President Maduro can't afford a lawyer for his trial in New York over alleged drug trafficking, Politico’s Kyle Cheney reported after reviewing recent court documents made public Wednesday. 

Maduro is hoping to use Venezuelan government money to fund his defense, but U.S. authorities refuse, citing sanctions against the country. Maduro’s current lawyer argues the block is effectively “interfering with Maduro’s Sixth Amendment right to counsel in perhaps the highest-profile criminal prosecution in the country,” Cheney writes. 

Elsewhere in the region: Mexico is considering legal action against Elon Musk after he alleged Monday that the president is linked to drug cartels and “saying what her cartel bosses tell her to say.” Pursuing such a case in the U.S. would be a longshot, however, because President Claudia “Sheinbaum would need to prove Musk knowingly said something false about her or recklessly disregarded the truth when he said it,” Reuters reported Tuesday. 

Additional reading: 

• “The Fall of a Strongman: Inside Maduro’s Last Days in Power,” via the New York Times, reporting Wednesday from Venezuela and Washington; 
• And “Violent Militias Stand Between the U.S. and Venezuela’s Vast Mineral Riches,” the Wall Street Journal reported Wednesday.