AI, Article II, and Immunity: What Could Go Wrong?
On Thursday, U.S. district judge Rita S. Lin issued a preliminary injunction against Defense Secretary Pete Hegseth’s declaration of Anthropic as a “supply chain risk,” finding that Anthropic would likely succeed on the merits that it was being retaliated against for protected First Amendment speech and that the Defense Department’s actions were arbitrary and capricious under the Administrative Procedures Act. (surprise!)
The order is worth reading in its entirety to fully understand how increasingly Micky Mouse this administration’s retaliatory efforts are becoming. Here is one paragraph:
At oral argument on Anthropic’s motion for a preliminary injunction on March 24, 2026, the Court inquired about the legal basis and effect of the following statement in the Hegseth Directive: ‘Effective immediately, no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic… . This decision is final.’ Counsel for Defendants agreed that he was not aware of any statute that gave Secretary Hegseth the authority to issue such a prohibition and agreed that the statement had ‘absolutely no legal effect at all.’ Counsel for Defendants also agreed that the statement did not reflect DoW’s ‘immediate intended course of action’ and that ‘DoW does not intend to terminate any contractors on the basis that they have a commercial relationship with Anthropic that [i]s separate from their work for DoW.’ Instead, counsel asserted that the only legal effect of the Hegseth Directive ‘flows through’ the Supply Chain Designation. When asked why Hegseth made a public statement that had no legal effect and that did not reflect the immediate intent of DoW, counsel stated, ‘I don’t know.’ Even so, Defendants declined to stipulate to enjoin this prohibition because DoW ‘is continuing to assess the situation’ and ‘will take action as needed … to mitigate the risk from Anthropic.’ (citations omitted and emphasis added)
Basically, at this point Hegseth, like other parts of this administration and Trump himself, are just making up their own “laws” that they think they can impose on anyone and everyone. And that is the rub.
A few weeks ago, I wrote about the ethical and moral issues at play with regard to Anthropic’s red lines in the use of Claude, its AI-powered technology, in the context of war. But the legal issues are equally important — and by that, I mean the lack of legal constraints or guardrails on how AI can be used. As you’ll recall, the dispute with Anthropic centers around its refusal to go along with a Defense Department contract that merely says that its technology will be used for “all lawful purposes” — Anthropic wanted two additional, specific exclusions stating that its technology would not be used for mass surveillance and for fully autonomous lethal weapons. I’m not sure whether Anthropic has a constitutional scholar on its payroll or just drew these red lines on its own, but if it’s the latter, it’s very smart.
You might think that “all lawful purposes” is a meaningful guardrail. But it’s important to remember what “lawful” means to this administration. As I have discussed previously on this Substack, this administration has a very expansive interpretation of Article II authority under the Constitution. This is especially so when it comes to the President’s Commander in Chief authority: As we have seen from the boat strikes on the high seas, the invasion of Venezuela, and now the war with Iran, this administration does not see any meaningful restrictions — even by Congress — on the ability to use military force. It also does not see any constraints under international law when it comes to legitimate military targets (known as distinction) or minimizing civilian casualties (known as proportionality). As Secretary Hegseth has clearly stated, these are just “stupid rules of engagement.”
Even worse is that when it comes to the use of military force outside the United States, the courts are unlikely to play any meaningful role, either. After President Obama ordered a drone strike on Anwar al-Awlaki, a U.S. citizen, in 2011, killing both him and his son (and another American), Awlaki’s family brought a wrongful death suit against the government. The judge dismissed the case as being outside the purview of the courts, stating:
In this delicate area of warmaking, national security, and foreign relations, the judiciary has an exceedingly limited role. This Court is not equipped to question, and does not make a finding concerning, Defendants’ actions in dealing with AQAP generally or Anwar Al- Aulaqi in particular….
[T]his case would impermissibly draw the Court into “the heart of executive and military planning and deliberation,” as the suit would require the Court to examine national security policy and the military chain of command as well as operational combat decisions regarding the designation of targets and how best to counter threats to the United States. Anwar Al-Aulaqi’s classification as a key AQAP leader to target by a drone strike raises fundamental questions regarding the conduct of armed conflict. The Constitution commits decision-making in this area to the President, as Commander in Chief, and to Congress. (citations omitted)
Great!
And to the extent that, somehow, someone would have standing to sue or some future prosecutor might bring charges against military officials who followed illegal orders, guess what? The Supreme Court has already handled that! After all, the Court’s immunity decision offers “absolute immunity” for any “official acts” taken pursuant to the President’s “core, exclusive, and preclusive” functions — and his Commander in Chief authority is one of them. While technically speaking, individuals under the president aren’t absolutely immune, in my opinion, they effectively are: Since you cannot use any official acts even as evidence of unlawful conduct by someone else, how would you bring a civil or criminal case against someone like Hegseth? And, of course, if all else fails the President has the, er, trump card: the pardon power.
What this means is that we (and the world) are largely at the mercy of AI platforms to stringently self-police, like Anthropic is doing. But Anthropic appears to be an outlier in this regard — OpenAI wasted no time swooping in to fill the void at the Defense Department. According to its public statement, OpenAI claims that it has created guardrails on the use of its technology, but there still seems to be a reliance on the “all lawful purposes” default, which, for the reasons above, are troubling.
“All lawful purposes” means little when you believe you’re above the law.
Your paid subscription makes the Freedom Academy possible — unique analysis based on my experience at the intersection of law, ethics, and national security. Plus, get access to guest speakers and book authors, and interactive discussions. Thank you for your support!