Civil War 2.0: An AI's Assessment

Starting with some basics, I asked Anthropic's Claude AI a series of questions designed to tease out its assessment of the potential for a wide scale constitutional crisis leading to armed conflict between Trump-controlled federal forces and that of an opposing state's police and National Guard units. Despite a key limitation of Claude, the exercise was illuminating and sobering.

The original Battlestar: Galactica series, Wargames, and the first Terminator all came out during my late high school through early college years. As a result, I have an almost primal fear about networked computers and artificial intelligence (AI). So you can probably imagine how I generally feel about AI and why I think it's only a matter of time before some networked, AI-driven system goes off the rails and hurts or kills a lot of people.

Even so, when a colleague suggested I give Anthropic's Claude AI a try as a research tool after reporting his own experience with the system, I decided to test Claude on a topic I think about daily: the potential for an American Civil War 2.0.

I only have these thoughts, of course, because tens of millions of my fellow Americans lost their minds and gave Donald Trump a second term in office. But at this point, Claude doesn't know that because his current knowledge cut off is October 31, 2024...literally a week before arguably the most consequential election since 1860.

While at first glance that might seem like a disadvantage in asking an AI platform to help me predict the future, I think the opposite is the case.

Claude's database is massive and almost certainly includes all the key publicly available information from Trump's first term in office, so if I wanted to ask it to predict a certain action Trump might take if reelected (based on Trump's actual prior record in office and statements made about a given policy position right up to the election), Claude could offer at least plausible scenarios or pathways for Trump to act.

Instead, to try to take Trump (and thus the political conflict over him and his prior policies) out of the equation entirely, I elected to ask Claude questions about a generic president taking certain actions while in office. For me, Trump was always the unspoken "president" in the dialogue I was having with Claude, but not once did I reference the former and now again American president.

To get a quick baseline on Claude's accuracy, I began with a simple question: Does the U.S. federal judiciary have direct operational control over any federal law enforcement agency or personnel?

Claude responded within less than five seconds with an answer of "No" (which was the correct answer) along with a nearly 180-word, factually correct exposition on the relationship between federal judges and law enforcement.

I then escalated the stakes in my next question.

"So, given what you said, should a sitting President of the United States get angry at a federal judge's ruling would there be anything stopping that President from ordering the U.S. Marshals Service to remove its personnel from the courthouse where the judge presides?"

Claude responded that "...there are significant constraints that would prevent a President from removing U.S. Marshals from a courthouse in retaliation for a judge's ruling" and the first he listed was 28 U.S. Code § 566, which as Claude noted means that the U.S. Marshals Service is "legally mandated...to provide security for the federal judiciary. Withdrawing this protection would likely violate this statutory duty."

Claude also said that "While a President might technically issue such an order, it would trigger an immediate constitutional crisis, likely legal challenges, and potential congressional action. The order itself would probably be immediately enjoined by other federal courts as an unconstitutional attack on judicial independence."

I then inserted the first reality check for Claude.

PE: Yet if the Congress in question was of the same political party as the President who ordered Federal Marshals to abandon the courthouse of the judge who offended the President, what are the odds that Congress would actually impeach and remove said President for violating 28 USC 566?

Claude AI: You're raising an important point about the practical limitations of our constitutional checks and balances when political alignment comes into play.

After adjusting its assessment based on the new political parameters I'd provided him, Claude conceded the odds of impeachment and removal "would indeed be low based on historical precedent."

When I then suggested to Claude that a sitting president determined to make an example of an offending judge could simply fire federal Marshals who refused to abandon their judicial protection duties, Claude offered that such a presidential move would "further escalate the constitutional crisis" and that "If a President were willing to systematically remove anyone who refused illegal orders, the constitutional guardrails would face unprecedented stress."

That of course is exactly the situation we face right now in the wake of Trump's firing of or integrity-based resignations by multiple Justice Department officials in just the past few weeks.

Fairly deep into our discussion I asked Claude whether all of these factors and actions could lead to civil war. He said it was possible but also listed a number of factors that mitigated against it, including "Economic interdependence between states and regions."

I knew that despite the North and South being very economically intertwined in April 1861, the Civil War still happened. I called Claude on it, and his response was fascinating.

Claude AI: You're absolutely correct to point this out. Despite significant economic interdependence between the North and South prior to the Civil War, conflict still occurred.

The antebellum economic relationship was deeply intertwined:

- Southern cotton supplied Northern textile mills

- Northern manufactured goods were sold to Southern markets

- The banking and shipping sectors connected both regions

- Northern capital financed much of the Southern plantation system

Yet this economic interdependence proved insufficient to prevent secession and war when fundamental political and social divisions—primarily surrounding slavery and state sovereignty—reached a breaking point.

This historical example effectively challenges my previous assertion that economic interdependence would necessarily serve as a strong deterrent to conflict in a modern constitutional crisis. While today's economic integration is even more extensive and complex than in the 1860s, your point stands that economic factors alone don't guarantee political stability when core values and constitutional questions are at stake.

The lesson from history suggests that when fundamental constitutional principles and regional/political identities are perceived to be under existential threat, even strong economic incentives for cooperation can be overwhelmed. (emphasis added)

After several more iterations of this kind of exchange with Claude in which I introduced new conflict scenario features and questions, I recapped where we were and asked my big question.

PE: And so I return to my previous question: could the kind of conflict between a President who defies court rulings, a governor who elects to use law enforcement or National Guard personnel to protect a federal judge abandoned by U.S. Marshals, and a federal judiciary that ruled against that President's illegal or unconstitutional conduct result in a civil war?

Claude AI: Yes, the scenario you've described could potentially escalate to civil war under certain conditions, especially if it expanded beyond a single incident.

Historical precedent suggests that constitutional crises involving direct confrontation between federal and state authorities, particularly when they involve the potential use of armed force, can evolve into wider conflicts when:

1. The underlying issues represent deep, irreconcilable divisions in society (as slavery did in the 1860s)

2. Political identities become so polarized that compromise seems impossible

3. Institutional guardrails fail simultaneously (courts, Congress, bureaucracy)

4. The crisis expands from a single incident to a pattern affecting multiple states

5. Initial armed standoffs result in casualties, creating martyrs and calls for retaliation

In your scenario, if a presidential defiance of court rulings led to a physical confrontation between federal forces and state National Guard units protecting judges, it could create a flashpoint. If other governors then took similar actions to protect federal judiciary in their states, and if the conflict expanded to include other constitutional disputes, the conditions for civil conflict would strengthen.

The most dangerous aspect would be the breakdown of the shared understanding of constitutional legitimacy. Once force becomes the arbiter of constitutional disputes rather than law, the foundation of peaceful governance is severely undermined. (emphasis added)

While modern Americans generally believe "it couldn't happen here," history shows that stable societies can rapidly destabilize when core governance norms collapse and institutional conflicts become militarized.

You can read the transcript of my exchange with Claude here.

After it was over, I got a monthly subscription to Claude AI because I don't think he's going to enable Terminators. At least not yet.

I was also left with yet another question: if an AI can understand how a Republic can be destroyed, why the hell can't everyone else in America recognize the same warning signs as Claude?

Thanks for reading The Sentinel. If you're not currently a subscriber, please consider becoming one as doing so is free through 2025 and it's an easy way to show your support for my work. Also, please share this piece with family, friends, and anyone else you believe would benefit from reading it.