‘Unable to determine probable cause’: Arizona judge releases DUI suspect over AI-generated police report

An Arizona judge has ordered the release of a 37-year-old man accused of driving under the influence after discovering that the arresting officer used an artificial intelligence tool to help prepare the police report.

Maricopa Superior Court Judge Daniel Fenzel said he could not establish probable cause because the report did not clarify which parts of the account had been written by the officer and which may have been generated by AI. Anthony Edward Rene Quidera was arrested on October 29 after an Arizona Department of Public Safety trooper allegedly spotted his car travelling at 100 miles per hour on Interstate 10 in Phoenix at around 12.33 am.

The arrest report stated that the trooper detected the smell of alcohol on Quidera's breath and found several open containers inside the vehicle. It also recorded that he displayed all six indicators of intoxication during field sobriety tests. However, the document included a disclaimer stating: "This report was generated with assistance from Truleo AI. The content reflects the officer's own observations and recollections," as cited by ABC News.

During Quidera's initial court appearance, Fenzel questioned the extent to which generative AI had contributed to the report. "The court is unable to meaningfully weigh whether the factual allegations made in the narrative were authored, whether in whole or in part, by a generative AI," the judge said. "Therefore, the court is unable to determine probable cause at this time."

Fenzel clarified that the decision did not prevent prosecutors from pursuing the DUI case at a later stage. Quidera was scheduled to appear for arraignment on Thursday. The Arizona Department of Public Safety confirmed that its officers use Truleo, an AI tool, to assist with drafting police reports. Raul Garcia, who represents the department, said its guidelines allow artificial intelligence to support the preparation of reports but require officers to remain responsible for the final document.

"Under Department policy, AI tools may be used only as an aid in preparing a report narrative. The trooper remains the author of the report and is responsible for its accuracy and completeness," Garcia said. He added that officers must check the draft against body-worn camera recordings and other available evidence before submitting it. They are also required to verify details such as names, timings, locations and actions, correct inaccuracies and obtain supervisory approval.

Garcia said troopers must complete department-approved training before using the technology. The case comes amid wider legal scrutiny of AI-generated material presented in court proceedings in Arizona. In a separate matter, the Arizona Court of Appeals ordered Gabriel Paul Horcasitas to be resentenced following his manslaughter conviction. The decision came after the victim's sister presented an AI-generated video depicting the deceased speaking during the sentencing hearing. In its September 30 ruling, the appeals court found that the trial judge's consideration of the video had prejudiced Horcasitas to such an extent that the sentencing process was fundamentally unfair.