The Robot Made Me Do It

Policing in the age of A.I.

Robocop (1987) depicts a dystopia in which a cyborg policeman struggles against his robot programming to recover his humanity… instead of the modern trend of the other way around.

On a balmy night of September 17^th, 2023, Jason Killinger strolled into the Peppermill Casino in Reno, Nevada. With cash in his pocket and a prayer on his lips, he hoped to hit the jackpot. Little did he know that the jackpot would hit him.

Killinger was hardly a high roller. By day he drove a delivery truck for the United Parcel Service. When time and money allowed, he played the strip in Reno, trying his luck at the various casinos. He would play low-stakes craps until he was up $200, then call it quits and move onto the next casino. That night he had already played the casinos of Legends Bay, the Nugget, and El Dorado. His hope was to play a while at the Peppermill – as he had hundreds of times before – then proceed to a fifth casino and call it a night.

Killinger never made it passed the Peppermill, however, for as he was playing video blackjack around 03:00, security staff approached him and addressed him as “Mike.” The staff demanded his identification, and despite Killinger’s current Nevadan driver’s license confirming that he was no Mike, staff insisted that he was. The security staff handcuffed and detained him, claiming that he was somebody named Michael Ellis.

In a cramped security office, Killinger discovered that the Peppermill Casino is equipped with what is purportedly a state-of-the-art security system which uses Artificial Intelligence (AI) through facial recognition. Multiple cameras alerted to Killinger’s presence on the casino floor as being a 99% match with one Michael Ellis, whom the Peppermill had trespassed about six months earlier for having fallen asleep on the casino floor. Security staff inventoried Killinger’s personal effects, amongst which were a valid Nevadan commercial driver’s license (CDL), a Peppermill player’s card, and a debit card, all in the name of Jason Killinger. Due to this incongruence, the staff called Reno Police Department, leaving the final determination to the responding officer.

Officer Richard Jager arrived at the Peppermill by 05:48. He reviewed the video footage and screenshots depicting Killinger on the casino floor hours earlier. He compared it to the archived images of Ellis from six months prior. You would be forgiven for thinking them to be one and the same person, for the likeness is uncanny, making it little wonder that the AI system flagged Killinger. That said, Officer Jager made a string of decisions during this incident, such that his career now dangles from that very string.

The Peppermill Casino’s security staff had scanned Ellis’ Nevadan driver’s license when he was trespassed, so Officer Jager called police dispatch to authenticate the license numbers of Ellis and Killinger, both of which the dispatch confirmed as legitimate. Officer Jager therefore announced his intention to arrest Killinger, as fingerprinting at the Washoe County jail would definitively determine his identity, absent any other evidence.

In a last-ditch effort to resolve the uncertainty, Killinger offered to produce additional personal documentation in his truck parked in the Peppermill’s lot. The truck contained registration, automotive insurance, a UPS paystub, a Teamster union card, and a medical card issued by the Department of Transportation, all in the name of Jason Killinger. But Officer Jager – who gushed about the Peppermill’s AI software as “pretty cool” – declined to review the contraindicative documentation so as not to impair the conclusion of guilt he had reached.

The Washoe County jail’s WIN check (Western Identification Network, a database shared amongst ten western states which cross-indexes criminal justice data with other governmental systems) determined within minutes that, surprise, surprise, Jason Killinger was not, and never had been, Michael Ellis. But like a common degenerate doubling down his last borrowed dollar at the craps table, Officer Jager continued his work in bad faith.

Officer Jager retconned the criminal complaint for trespass – originally drafted in the name of John Doe – to the name of Jason Killinger. He attested in his arrest report that the only disparity between Killinger and Ellis was their listed eye color (Killinger’s being blue and Ellis’ hazel, which admittedly is close), but Officer Jager crucially omitted several others. There was a difference in age between Killinger and Ellis of seven years, with Killinger the junior at forty years old on that night. There was a difference in the recorded weights, with Killinger listed as fifty pounds heavier than Ellis. The signatures were markedly different, and perhaps most importantly, Killinger had a valid CDL issued by Nevada, printed on a Real ID card, which per se meant that two sovereign governments, both state and national, had been assured of his identity.

Thankfully some assistant district attorney realized the absurdity of charging Killinger for trespass when that charge would only have applied had Michael Ellis returned to the Peppermill Casino, so the charges were dropped. That still leaves Killinger with a criminal record which will come up in background checks, and injured shoulders from having been kept in handcuffs for so long, despite his repeated complaints of acute discomfort. And this ordeal could have been avoided, if only Officer Jager had reviewed Killinger’s additional documentation on the scene.

Officer Jager’s connivance to avoid countervailing documentation of Killinger’s identity mirrors another case. One of the first cases students read when studying criminal law is United States v. Jewell, 532 F.2d 697 (1976). This case involved a guy named Jewell who agreed to a known drug dealer’s proposition that Jewell drive a truck from Tijuana, Mexico, to the USA. Jewell hopped right in, not bothering to investigate a conspicuous compartment which, unsurprisingly, contained 110 pounds of marijuana, as U.S. Border Patrol happily uncovered. The case stands for the principle that for those crimes which require the criminal’s knowledge of his wrongdoing, he cannot avoid guilt by actively ignoring hints about that wrongdoing.

If such a principle is good enough to apply to the citizenry in its dealings with government, then conversely it must apply to the government in its dealings with the citizenry. Stated specifically, if review of Killinger’s additional documentation within his on-site vehicle could have resolved the dispute about his identity, then Officer Jager must be liable for purposely disregarding it when he decided to arrest Killinger and weaponize the criminal justice system against him.

Jason Killinger originally would have been happy that night to leave the Peppermill Casino with gains of just $200. As things now stand, he has a good case against the casino, Reno Police Department, the AI software’s manufacturer, and Officer Jager individually, if Jager doesn’t cower behind the skirt of qualified immunity. For Killinger, this may mean a sizable jackpot in judgments against these parties.

What Happens in Vegas, Never Stays in Vegas

Granted the casino town in question was Reno, not Las Vegas, but the idea’s still the same. There is no way that a minor arrest in Nevada has no implications for the country as a whole. The current administration is joined at the hip with Palantir, a shadowy mass AI surveillance grid run by tech brah and major Trump-donor Peter Thiel (mainstream press frequently applies to Thiel the epithet of “libertarian,” which only shows that the press has no clue what the term means!). Cities and towns across the United States are rapidly installing surveillance systems operated by AI, like the infamous Flock cameras. As proprietary technology, there is no telling the extent of what they capture, to whom or what the data is siphoned, and how it is used.

The blackbox nature of this technology was spectacularly demonstrated for the whole country during Ring® camera’s commercial during the Super Bowl LX on February 8^th, 2026. The commercial revealed that any rando can query footage from any Ring camera user, though supposedly only to find lost dogs. This was a feature to which many Ring customers did not know they were subscribed, since reportedly their consent was presumed by default. Thankfully many Americans realized that it would be a short stretch for Ring’s cameras to start surreptitiously monitoring human beings. Thus, some customers have been removing the cameras, not wanting their front doors to be used as unwitting nodes in a potentially global spy network like that “foreshadowed” in Christopher Nolan’s dystopian fiction The Dark Knight (2008) and made sexy by Batman himself.

Another worrisome instance of doorbell camera footage being archived against the manufacturer’s assurances to the contrary was recently demonstrated by the TODAY Show host Savannah Guthrie’s mother Nancy’s abduction from her home in Tucson, Arizona. Nancy Guthrie had installed a Google Nest® doorbell camera, but had not subscribed to remote archiving or monitoring, meaning that she could only see whatever was in front of her doorbell cam at time of viewing. Tucson Police lamented that no video recording existed which may have captured Guthrie’s abductor, but the victim as mother to a celebrity caused too much uproar, and the Federal Bureau of Investigations’ Director Kash Patel had to confess that all footage is secretly archived and retrievable by government… but only if you’re famous enough to be worth the effort.

We do know for certain just two things. Firstly, whereas Flock is premised on the notion that this dragnet can capture any person or thing in public for government’s review, the citizen will always have to jump through hurdles like refusable FOIA requests to retrieve captures about government which were supposed to be “public,” such as which police officers were hanging out late at the local cop bar on the night when some neighbor’s car parked streetside was side-swiped by a drunken hit-and-run. Secondly, as Jason Killinger’s case aptly proves, we know that law enforcement is not yet trained to the task of investigating AI’s false positives.

In a conflict between identifications, when a fancy AI system was asserting one identity, and significant evidence at the scene was asserting another, Officer Jager’s one job was to evaluate the analog evidence at hand. Though his overall demeanor was personable and he never insulted Killinger or anything of that sort – composure which seems rare in the genre of police bodycam videos – he connived a case against a citizen he had sworn to protect, and only to make a computer look “cool.” Sergeant Jamie Milliman of Columbine Police Department in Colorado similarly skipped diligent investigation to leap straight to conclusions of guilt about a woman pilfering a porch package worth $25 then escaping in her Rivian EV truck worth $76,000. Police in Jacksonville, Florida, used AI to obtain an arrest warrant for Robert Dylan, being that his face was a 93% match with that of a suspected child molester, when the most which the law authorized was for police to use AI as an investigatory lead to be followed up by questioning and other sleuthing.

Our society is still a ways from the effective and ethical use of AI technology. There will be growing pains, as there were with every major innovation. Until we have a surer idea of how to deal with misidentifications and such, there is no better substitute for the police officer’s thorough investigation of analog evidence… and a smattering of common decency.