Arizona court tosses sentence over AI video of a dead victim
The Arizona Court of Appeals vacated a sentence after an AI-generated video of the victim swayed the judge. What it means for AI likenesses in your content.
The Arizona Court of Appeals has vacated a 10½-year manslaughter sentence because an AI-generated video of the victim unfairly swayed the trial judge. It happened in our backyard: a Chandler road-rage case, sentenced in Maricopa County. The ruling is about a courtroom, but the lesson travels. An AI video of a real person is not evidence of what that person thinks.
What actually happened
- Gabriel Horcasitas was convicted of manslaughter in the November 2021 road-rage shooting of Christopher Pelkey.
- At the May 2025 sentencing, Pelkey's sister played a video she made with AI. It animated a photo of her brother, who "spoke" a script she wrote, including words of forgiveness.
- Maricopa County Superior Court Judge Todd Lang said he "loved that AI" and gave the maximum, 10½ years.
- The appeals panel found the video showed the victim's thoughts as created from his sister's "imaginings," not real events, and that it rendered the sentencing "fundamentally unfair."
- The conviction stands. The sentence is vacated and the case goes back for resentencing.
The court did not ban AI victim statements in general. It said this one, used this way, crossed the line.
Why it matters for your business
A synthetic person is a script with a face. The court's core point is simple: the words came from the person who wrote the prompt, not from the person on screen. The same is true for an AI "testimonial," an AI version of your founder, or a cloned spokesperson in an ad. If the real person did not say it, the video does not prove they believe it.
Label it, and get consent in writing. If you use AI likenesses in marketing, training videos or support, say so on screen. Use only people who signed a release that covers AI generation, and keep that release on file. Never animate a customer, an employee who left, or anyone who cannot approve the script.
Expect the rules to tighten. Courts, regulators and platforms are all writing policy on synthetic media now. Content that is clearly labeled and consented is easy to defend. Content that blurs the line is the next lawsuit.
Key takeaways
- The Arizona Court of Appeals vacated a 10½-year sentence over an AI video of the victim
- The panel said the video showed the sister's "imaginings," not the victim's real views
- The conviction stands; resentencing follows. AI victim statements are not banned outright
- Treat any AI likeness as a script with a face, never as proof of what a person thinks
- Label AI likenesses on screen and get written consent that covers AI generation
Using AI video or voice in your marketing? We build AI content workflows with disclosure labels, consent tracking and an approval step a human signs off on. See how we build or talk to our Phoenix team.
- #ai-video
- #ai-likeness
- #arizona
- #ai-disclosure
- #deepfakes
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Fake ChatGPT Custom GPT ads push ClickFix RAT malware
Huntress found Google ads for 'chatgpt' leading to a malicious Custom GPT on chatgpt.com that ends in a ClickFix RAT. Bookmark your AI tools; stop searching.
Read itCloudSyncD: fake Zoom installer backdoors Macs via your password
Jamf found a fake Zoom installer that asks for your Mac password, then runs the CloudSyncD backdoor with sudo. Here is how to stop it on your team's Macs.
Read it