U.S. Marshals arrested Angela Lipps, 50, at her home in Elizabethton, Tennessee, on July 14, 2025. She was babysitting four children at the time. Let me say at the outset what should not need saying in a nation that reveres the Fourth Amendment: it does not matter that she is a grandmother. It would not matter if she were a college student, a felon, or a stranger to any of us. What matters is that the government took her liberty on the word of a machine, and no one along the chain of custody that followed thought it necessary to verify that word before acting on it.
The test of a constitutional system is not how it treats the sympathetic defendant. It is how it treats everyone else. Angela Lipps happens to be sympathetic. That is a coincidence, not a legal argument, and it should be set aside so the actual issue can come into focus.
Here is what happened, stripped of sentiment. West Fargo police ran a photograph, taken from a fake military identification used in a bank fraud scheme, through a private facial recognition system called Clearview AI, a system West Fargo had purchased on its own and never disclosed to the command staff of the neighboring Fargo Police Department. Clearview returned Lipps as a “potential match.” A Fargo detective compared that unverified computer guess to her driver’s license photo and her social media accounts, decided it was sufficient, and sought a warrant. On July 1, 2025, a North Dakota judge signed that warrant and authorized nationwide extradition. Nobody, at any point in this sequence, called Tennessee to ask where she had been in April and May of that year. Her bank records, which would eventually end the case in a single afternoon, sat untouched the entire time.
It is worth being precise about why this matters as a matter of law, because the popular outrage, however justified, is aimed at the wrong target. The outrage is aimed at the algorithm. It should be aimed at the human beings who treated the algorithm’s guess as a substitute for their own constitutional obligations.
Understand what a warrant is supposed to represent. It is supposed to represent an independent judicial determination, based on the totality of the circumstances, that probable cause exists. That is the rule of Illinois v. Gates, and it has been the rule for over four decades. The presence of a judge’s signature is not decorative. It exists so that the coercive power of the state cannot be deployed on an officer’s say so alone.
But a judge’s signature is not a magic shield. It is a rebuttable presumption. If the officer who sought that warrant knew, or should have known, that the underlying identification was an unverified computer guess rather than a corroborated fact, and did not say so in the affidavit, then the judge was not exercising independent judgment. The judge was rubber stamping a conclusion the officer had already reached. That is precisely the defect the Supreme Court identified in Franks v. Delaware, and it is precisely why Malley v. Briggs holds that qualified immunity evaporates when no reasonably competent officer could have believed the affidavit supported probable cause. It is not yet known, because no complaint has been filed, whether Lipps’s warrant application disclosed that Clearview’s own guidance treats its results as investigative leads requiring independent corroboration, not probable cause in themselves. If it did not disclose that, the case is not close.
The Detroit precedent is instructive, and coincidences of this magnitude are rare enough to warrant attention. Robert Williams was arrested in 2020 on a facial recognition match that no one corroborated. He settled his case against the city of Detroit for $300,000 in 2024, on the theory that the detective’s warrant application concealed from the magistrate exactly how unreliable that match actually was. If the pattern holds here, and everything reported so far suggests it does, then Fargo is not looking at an unfortunate technological error. It is looking at a repeat of a legal theory that has already proven itself in federal court.
The individuals most responsible for the charging decision are likely protected by absolute immunity under Imbler v. Pachtman, a doctrine that shields prosecutorial conduct more broadly than many would find comfortable, and that remains the law regardless. There is a narrow exception, articulated in Kalina v. Fletcher, for a prosecutor who personally swears to facts in a probable cause affidavit rather than simply exercising professional judgment. Whether that exception applies here depends on documents the public has not yet seen. The judge who signed the warrant enjoys immunity that is close to absolute regardless of what the affidavit contained. That leaves the detective, who may be exposed under Franks and Malley, and it leaves the municipalities themselves, which can be held liable under Monell v. Department of Social Services where the violation flows from an actual policy failure rather than one officer’s individual lapse. An agency that allowed an uncertified private vendor to operate without its own command staff’s knowledge, and that then assumed, wrongly, that proper verification had occurred elsewhere, is not describing a rogue actor. It is describing a policy failure, and policy failures are exactly what Monell was designed to reach.
The state’s attorney reviewed the file, a judge found probable cause, a second judge authorized extradition. Three institutional actors, the argument goes, cannot all have failed. But three institutional actors relying sequentially on the same uncorroborated tip is not three independent judgments. It is one judgment, repeated three times by people who each assumed someone earlier in the chain had done the actual verifying.
The Supreme Court held in Baker v. McCollan that a few days of detention under a facially valid warrant does not itself violate due process, even in the face of repeated protests of innocence, because the Fourth Amendment’s probable cause requirement is doing the constitutional work at that stage. But the Court explicitly declined to say what happens when a few days becomes many months, when the exculpatory evidence exists the entire time in a bank’s own records, and when no one looks until a public defender is finally assigned the file. Lipps’s case sits in exactly that unresolved space, and it would be a plausible vehicle for testing it.
As for why her Tennessee alibi did not free her sooner, the answer lies in extradition law, and it is an answer that will frustrate anyone expecting common sense to prevail quickly. Under Michigan v. Doran, once the demanding state’s judicial officer has found probable cause and the governor has acted, a court in the state where the fugitive is held may review almost nothing beyond the sufficiency of the paperwork and the identity of the person named. It cannot relitigate guilt. This is why her release came only when a North Dakota public defender handed her bank records to a North Dakota prosecutor, months after her arrest, rather than through any earlier intervention by a Tennessee court. Understanding why the system moved so slowly is the only way to know where reform actually belongs.
A system of justice is not measured by how it treats the cases that resolve themselves easily. It is measured by how it treats the cases where a machine’s uncorroborated guess travels, unchallenged, through a detective, a prosecutor, and two judges, and ends only when a woman personally suffers five months of incarceration for a crime that a bank statement could have disproven on day one. Whether Angela Lipps ultimately prevails in court will depend on facts not yet public. Whether she should prevail is not, on the present record, a difficult question. It rarely is, once the algorithm is taken out of the sentence and the human beings around it are asked what they actually did.