Arizona judge frees DUI suspect after spotting AI-written police report
A disclaimer flagging AI assistance in an arrest report left a judge unable to tell what the officer actually witnessed, so he released the suspect pending further review.
What happened: Maricopa County Superior Court Judge Daniel Fenzel released Anthony Quidera, 37, who was accused of driving 100 mph while intoxicated on a Phoenix interstate, after noticing his arrest report ended with a disclaimer saying it was written with help from an AI tool called Truleo. The trooper's report described smelling alcohol, finding open containers, and Quidera failing all six field sobriety cues. But Fenzel said he could not tell how much of that narrative came from the officer versus the software, so he ruled he could not determine probable cause for now. The case isn't dismissed; an arraignment was still scheduled.
Why it matters: Probable cause hearings depend on judges trusting that an officer's written account reflects their own firsthand observations. When a report discloses AI involvement but gives no way to see what was added, edited, or phrased by software, that trust becomes harder to verify. This is one of the first known cases where an AI disclaimer alone was enough to stall a prosecution, and it signals that courts are starting to treat AI-assisted paperwork as a question mark rather than a routine fact.
How it works, plainly: Arizona's Department of Public Safety uses Truleo AI to help troopers draft report narratives, often from notes, dictation, or body-camera audio. Department policy says the AI only assists; the officer is still the named author and must check the draft against body-camera footage, fix any errors in names, times, or actions, and get a supervisor's sign-off before it becomes official. Officers must also complete training before using the tool. What the policy doesn't produce is a visible record of which sentences were AI-generated versus human-written, which is the gap the judge flagged.
The rollout: This case lands weeks after an Arizona appeals court ordered a new sentencing hearing for a different man, ruling that an AI-generated video showing a deceased victim speaking to the court had unfairly swayed the judge. Together, the two rulings suggest Arizona courts are actively working out how much weight AI-touched evidence and documents should carry, likely pushing other departments and courts to add clearer labeling or stricter review rules before similar tools spread further into daily casework.