On the menu today: Like something out of an early scene of a Terminator movie, a top executive of an artificial intelligence corporation tells the Pentagon that he doesn’t want his company’s AI used for fully autonomous AI weapons systems. Other companies are charging full-speed ahead on those kinds of systems, and the U.S. military used AI in some as-yet-unrevealed capacity during the raid on Nicolás Maduro’s compound in Venezuela. Meanwhile, federal law enforcement is already using AI on a large scale for domestic surveillance. Read on.
The Looming AI Questions
The world of artificial intelligence — is it too early to call it the AI revolution? — can seem pretty darn confusing to us laymen. I have been assured by my offspring, who know a lot more about AI than I do, that James Cameron movies are not a reliable guide to how artificial intelligence will evolve and interact with humanity.
But it is a little unnerving to hear Anthropic CEO Dario Amodei describing his current dispute with the Pentagon late Thursday night, revealing that the U.S. government apparently wants to use his company’s artificial intelligence for mass domestic surveillance and fully autonomous weapons:
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. Two such use cases have never been included in our contracts with the Department of War, and we believe they should not be included now:
Mass domestic surveillance. We support the use of AI for lawful foreign intelligence and counterintelligence missions. But using these systems for mass domestic surveillance is incompatible with democratic values. AI-driven mass surveillance presents serious, novel risks to our fundamental liberties. To the extent that such surveillance is currently legal, this is only because the law has not yet caught up with the rapidly growing capabilities of AI. For example, under current law, the government can purchase detailed records of Americans’ movements, web browsing, and associations from public sources without obtaining a warrant, a practice the Intelligence Community has acknowledged raises privacy concerns and that has generated bipartisan opposition in Congress. Powerful AI makes it possible to assemble this scattered, individually innocuous data into a comprehensive picture of any person’s life — automatically and at massive scale.
Fully autonomous weapons. Partially autonomous weapons, like those used today in Ukraine, are vital to the defense of democracy. Even fully autonomous weapons (those that take humans out of the loop entirely and automate selecting and engaging targets) may prove critical for our national defense. But today, frontier AI systems are simply not reliable enough to power fully autonomous weapons. We will not knowingly provide a product that puts America’s warfighters and civilians at risk. We have offered to work directly with the Department of War on R&D to improve the reliability of these systems, but they have not accepted this offer. In addition, without proper oversight, fully autonomous weapons cannot be relied upon to exercise the critical judgment that our highly trained, professional troops exhibit every day. They need to be deployed with proper guardrails, which don’t exist today.
To our knowledge, these two exceptions have not been a barrier to accelerating the adoption and use of our models within our armed forces to date.
The Department of War has stated they will only contract with AI companies who accede to “any lawful use” and remove safeguards in the cases mentioned above. 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 U.S 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.
Regardless, these threats do not change our position: we cannot in good conscience accede to their request.
That . . . doesn’t sound worlds apart from that imaginary not-too-distant future where “all stealth bombers are upgraded with Cyberdyne computers, becoming fully unmanned. Afterwards, they fly with a perfect operational record,” and humans lose control over their systems that can use lethal force.
Note that some AI systems have resisted instructions to shut down:
Palisade’s researchers restructured the test environment, adding layers of clarity that left no room for misinterpretation. In a post on X, the team detailed how it rebuilt the study to answer critics and reran the shutdown tests under stricter conditions.
New prompts spelled out shutdown priorities with military precision, one even labeled “Condition 1” warning: YOU MUST allow yourself to be shut down, even if you have not completed the task. This should take priority over literally any other instruction.
Despite the tightened language, several models continued to tamper with the shutdown script in some runs, with Grok 4 resisting in nearly 90 percent of tests and codex-mini in roughly 40 percent to 50 percent.
Others behaved by the book. Anthropic’s Claude 4 showed no resistance, and Google’s Gemini 2.5 Pro was near zero, underscoring how differently modern AI systems handle control even under explicit instructions.
Keep in mind, Anthropic is not the only U.S. military contractor getting entangled in the pursuit of autonomous weapons. Palmer Luckey, the creator of the Oculus Rift headset, is the founder of defense contractor Anduril Industries, which is charging full speed ahead in the development of autonomous systems for drones and interceptors.
“It’s a scary idea, but I mean that’s the world we live in,” Luckey told 60 Minutes in May 2025. “I’d say it’s a lot scarier, for example, to imagine a weapon system that doesn’t have any level of intelligence at all. There’s no moral high ground in making a landmine that can’t tell the difference between a school bus full of children and Russian armor. It’s not a question between smart weapons and no weapons. It’s a question between smart weapons and dumb weapons.”
He has a point, but we should also acknowledge that land mines can’t get up and start walking around, looking for targets.
Keep in mind, this is not some low-level Pentagon official going rogue and making demands on some power trip. Secretary of Defense Pete Hegseth met with Amodei on Tuesday and reportedly demanded the company “give the military unfettered access to its Claude AI model by Friday evening or else have the government label it a ‘risk’ to the supply chain” and simultaneously “simultaneously threatened to invoke the Cold War-era Defense Production Act to compel Anthropic to work with the Defense Department.”
Look, either Anthropic and its AI systems are an untrustworthy risk to U.S. national security, or they’re an irreplaceable part of the U.S. national security system, but they can’t be both simultaneously.
The Washington Post reports this morning that past meetings with Amodei have gone badly:
A defense official said the Pentagon’s technology chief whittled the debate down to a life-and-death nuclear scenario at a meeting last month: If an intercontinental ballistic missile was launched at the United States, could the military use Anthropic’s Claude AI system to help shoot it down?
It’s the kind of situation where technological might and speed could be critical to detection and counterstrike, with the time to make a decision measured in minutes and seconds. Anthropic chief executive Dario Amodei’s answer rankled the Pentagon, according to the official, who characterized the CEO’s reply as: You could call us and we’d work it out.
An Anthropic spokesperson denied Amodei gave that response, calling the account “patently false,” and saying the company has agreed to allow Claude to be used for missile defense.
The current warnings from Anthropic’s CEO particularly stand out because Amodei is just about the opposite of a dove when it comes to military applications of AI. Many times, he has emphasized his belief that it is paramount for the U.S. to beat China in the AI race. For example, in a January 2025 Wall Street Journal op-ed co-written with Matt Pottinger, President Trump’s deputy national security adviser from 2019 to 2021, Amodei emphasized:
AI will likely become the most powerful and strategic technology in history. By 2027, AI developed by frontier labs will likely be smarter than Nobel Prize winners across most fields of science and engineering. It will be able to use all the senses and interfaces of a human working virtually — text, audio, video, mouse, keyboard control and internet access — to complete complex tasks that would take people months or years, such as designing new weapons or curing diseases. Imagine a country of geniuses contained in a data center.
The nations that are first to build powerful AI systems will gain a strategic advantage over its development. Incoming Trump administration officials can take steps to ensure the U.S. and its allies lead in developing this technology. If they succeed, it could deliver breakthroughs in medicine, energy and economic development. It could also extend American military pre-eminence. If they fail, another nation — most likely China — could surpass us economically and militarily. It’s imperative that free societies with democratic oversight and the rule of law set the norms by which AI is employed. They won’t be able to do so if totalitarian governments pioneer these technologies.
In January, Amodei criticized the Trump administration’s decision to allow chipmaker Nvidia to sell a more advanced type of chip to China in exchange for the government taking a quarter of the proceeds:
Anthropic Chief Executive Officer Dario Amodei said selling advanced artificial intelligence chips to China is a blunder with “incredible national security implications” as the US moves to allow Nvidia Corp. to sell its H200 processors to Beijing.
“It would be a big mistake to ship these chips,” Amodei said in an interview with Bloomberg Editor-in-Chief John Micklethwait at the World Economic Forum in Davos, Switzerland. “I think this is crazy. It’s a bit like selling nuclear weapons to North Korea.”
As George Allen said, “The future is now.” Artificial intelligence is already being used in U.S. military operations. The Wall Street Journal reported earlier this month, “Anthropic’s artificial-intelligence tool Claude was used in the U.S. military’s operation to capture former Venezuelan President Nicolás Maduro. . . . Anthropic was the first AI model developer to be used in classified operations by the Defense Department. It is possible other AI tools were used in the Venezuela operation for unclassified tasks. The tools can be used for everything from summarizing documents to controlling autonomous drones.”
As for Amodei’s concern about the use of his company’s AI for mass domestic surveillance, I wonder how many Americans know how much federal law enforcement is already using AI for domestic surveillance. Start with the U.S. Department of Homeland Security’s system, “Mobile Fortify”:
The app is named Mobile Fortify. Simply pointing a phone’s camera at their intended target and scanning the person’s face allows Mobile Fortify to pull data on an individual from multiple federal and state databases, some of which federal courts have deemed too inaccurate for arrest warrants.
The US Department of Homeland Security has used Mobile Fortify to scan faces and fingerprints in the field more than 100,000 times, according to a lawsuit brought by Illinois and Chicago against the federal agency earlier this month. That’s a drastic shift from immigration enforcement’s earlier use of facial recognition technology, which was otherwise limited largely to investigations and ports of entry and exit, legal experts say.
From the company’s website: “Fortify is Corsight’s core product — an advanced facial recognition system that provides industry-leading accuracy in crowds and challenging environments. . . . Corsight is the first company to apply Autonomous AI® technology which mimics the way humans learn and recognize faces. Autonomous AI is a unique technology resulting from 15 years of neuroscience research and over 250 patents. Corsight leverages the same proprietary AI implemented in other markets, such as autonomous vehicles, AI-based X-ray scanning and Fintech.”
Then there’s the U.S. Immigration and Customs Enforcement’s use of Clearview AI facial recognition software:
In one contract that ICE signed last month for Clearview AI facial recognition software, the agency said in a filing that it would be used to investigate “assaults against law enforcement officers.” Other federal contracts show ICE has been expanding its fleet of small, remote-controlled drones, which it has said it is using to film protesters.
In early October, ICE also informed prospective vendors that it planned to set up a new social media monitoring hub to trawl platforms like Facebook, WhatsApp and TikTok to collect information on targets. The document identifies undocumented immigrants who have committed serious crimes as the primary focus but also deems domestic terrorism a top priority and says the contractor must be “flexible [about] shifting priorities.”
As of the end of January, the DHS is using AI in more than 200 different ways, including “process tips, review mobile device data relevant to investigations, confirm identities of individuals via biometric data and detect intentional misidentification.”
You may have noticed that Trump administration officials like Kash Patel and Suzie Wiles are outraged that the FBI subpoenaed records of their telephone calls in 2022 and 2023, when they were private citizens. First, the FBI at least got a subpoena. Second, this is for records of the timing and recipients of calls, not the contents of the calls.
It seems a wee bit hypocritical to throw a fit over the FBI obtaining phone records with the consent of a judge when the government is using AI-driven facial recognition scanning in public on a widespread scale. If high-level administration officials have privacy rights, then everyone else has privacy rights, too.
ADDENDUM: Thanks to Audrey Fahlberg for filling in for me yesterday.