Arizona Court Overturns Manslaughter Sentence After Judge Relied on AI-Generated Video of Victim

Arizona Court Rules AI-Generated Victim Impact Statement Violated Due Process

The Arizona Court of Appeals has ordered resentencing in a manslaughter case after determining that an AI-generated video depicting deceased victim Christopher Pelkey speaking to the court was unreliable evidence1

1

. Gabriel Paul Horcasitas, who received a 10-year manslaughter sentence for fatally shooting Pelkey during a November 2021 road rage incident in Chandler, Phoenix, Arizona, must now face resentencing3

3

.

The three-judge panel ruled Wednesday that the AI-generated victim impact statement crossed a fundamental line. "Rather than document an event or recording a particular moment, the AI video presents a depiction of the victim and his thoughts created from the imaginings of the victim's sister," the court wrote1

1

. This marks what legal experts believe is the first time a U.S. appellate court has overturned a sentence based on AI-generated evidence presented during sentencing.

First-of-Its-Kind AI-Generated Victim Statement Raises Admissibility Concerns

Pelkey's sister, Stacey Wales, created the synthetic likeness and voice of her brother after struggling to determine what he would say2

2

. The AI rendering showed Pelkey expressing forgiveness and suggesting that in another life, he and Horcasitas "probably could have been friends"4

4

. The video was played during a May 2025 sentencing hearing after nine family members and friends had already testified1

1

.

Superior Court Judge Todd Lang commented that he "loved that AI" and called it "genuine," noting Pelkey's "obvious forgiveness of Mr. Horcasitas reflects the character I heard about today"2

2

. The judge handed down a punishment exceeding what prosecutors had requested, though he didn't explicitly state whether the AI-generated video factored into his decision2

2

.

Defense Argues Fundamental Error in Relying on Unreliable Evidence

Kristen Reller, Horcasitas' appellate lawyer, argued that her client had no meaningful opportunity to rebut material in the AI-generated video and that Judge Lang violated due process protections by relying on AI evidence1

1

. Reller contended the video carried "undue emotional weight" and conveyed an authenticity that wouldn't have existed had a family member simply read the same words aloud3

3

.

The defense also highlighted that the video didn't disclose who wrote the words used by the AI version of Pelkey and wasn't backed up with evidence establishing that its contents accurately reflected Pelkey's actual views3

3

. The Maricopa County Attorney's Office stated that prosecutors knew the victim's family would address the court during sentencing but weren't aware of the nature of the presentation1

1

.

Victim's Family Expresses Disappointment Over Ruling

Stacey Wales expressed disappointment that the AI-generated video was cited as the reason for the appellate court's decision to overturn the sentence tossed in Arizona case. "It feels unfair because there is a convicted murderer sitting in prison that has blankets and walls of protection around their rights. Where are the rights for the victim?" Wales said2

2

. She stated her only goal was to humanize her brother for the judge3

3

.

A victim appeals lawyer has informed the family that using AI again could result in another appeal. Wales responded: "We will continue to let his voice be heard in whatever allowable medium we can convey that through the court system"3

3

.

Implications for AI Use in Judicial Proceedings

While AI use in judicial proceedings has expanded, it's typically been reserved for administrative tasks, legal research and case preparation1

1

. In Arizona, AI has helped inform the public of rulings in significant cases. However, using AI to generate victim impact statements marks a new frontier for sharing information with the court outside evidentiary phases4

4

.

This ruling raises critical questions about the admissibility of evidence created through synthetic media. Courts will need to establish clear guidelines on when AI-generated content can be presented during sentencing and what safeguards must be in place to protect due process rights. The emotional weight of seeing and hearing a deceased victim speak—even through AI—presents unique challenges for maintaining fairness in judicial proceedings.

Legal experts will be watching closely as Horcasitas faces resentencing. The case sets a precedent that could influence how courts across the United States handle AI-generated evidence, particularly in emotionally charged victim impact statements where the fundamental error of relying on unreliable evidence could prejudice judicial decisions.