Christopher Pelkey was dead when a video of him addressed the man convicted of killing him.

His family had recreated him using AI. The judge knew that, yet described the forgiveness he heard as genuine.

On September 30, an Arizona appeals court threw out Gabriel Paul Horcasitas's manslaughter sentence and ordered resentencing. His conviction remains intact. Court ruling

The detail I keep coming back to is that nobody needed to mistake the video for a real recording. Knowing how it was made still left room to treat its message as Pelkey's own.

I am Alex. Welcome to ShortCu8 by Innov8.

Lets Dive Deep.

Before trusting a familiar face or voice, find out who supplied the words.

A recording of something a person said, a script they approved and somebody else's idea of what they might say deserve different treatment.

Pelkey was killed during a road-rage confrontation in 2021. His sister, Stacey Wales, helped create the video presented at the sentencing hearing in May 2025. AP

It is easy to understand the wish behind that. When someone you love is gone, you still imagine their reaction to things. You remember their humour, beliefs and habits. Sometimes you can almost hear the answer.

Making that imagined answer visible introduces another choice: letting an audience hear it through the person's recreated face and voice.

I would want to keep the family’s grief in view here. This is a difficult boundary precisely because the intention can be loving.

The recreation disclosed that it used AI. It also included actual footage of Pelkey. Judge Todd Lang praised the depicted forgiveness, treating it as an expression of the victim's character. Opinion, paragraphs 36 and 41

The appeals court found that the invented depiction presented the family's interpretation as coming directly from the victim. It said the video played a role in sentencing, while acknowledging that its precise effect couldn't be determined. Opinion, paragraphs 50–53

That raises a question worth carrying beyond the courtroom. Can we know a face is generated and still give its words the weight we would give the person?

This case doesn't answer that for everyone. It does show why an AI label alone may leave the more important question unanswered: whose views are being expressed?

In our video-calling issue, we looked at increasingly convincing AI conversations. This story adds a question about authorship.

Imagine a founder approving an avatar to read a product introduction. Later, somebody gives it a new script announcing a refund policy. The face stays familiar, but approval of the first video tells us nothing about the second.

Or imagine a family creating a memorial message. Showing an old recording preserves something the person actually said. Writing a new message involves an interpretation, however thoughtful it is.

For me, the useful boundary is approval of the specific message. Permission to recreate a face should come with a clear understanding of what it will say and where it will appear.

When a recreated person delivers an important message, ask who wrote it and whether the person approved those words or actually recorded them. If the answer is unclear, verify it before acting on the message.

If you're making the video, explain the authorship as well as the use of AI. A label such as “AI recreation; new script written by the family” tells viewers more than “made with AI.” That still doesn't establish that the person would agree with the script, or make it suitable for every setting.

And if you're creating your own avatar, decide whether each new script needs your approval. Put that boundary in place while you can still choose it yourself.

The court upheld the conviction and required a new sentencing hearing. Its decision concerns this presentation and its effect on fairness, rather than establishing a blanket ban on AI memorials. Judgment

The advice about avatars above is my interpretation of the authorship problem, rather than a legal rule from the case.

I would want my family to keep the recordings where I chose my own words. Before giving anyone a version of me that can say new things, I would want us to agree on who gets to write them.

We read every reply - just reply to this email and let us know how we can improve !

#AD1

If you have paid time left on an individual Granola plan, we'll match it with a Wispr Flow subscription that includes Notetaker and dictation, and add bonus time, up to 12 months total. Sign in or create a Wispr account and submit proof of your plan to check eligibility.

Try Notetaker Free

#AD2

The International Quant Championship (IQC) is one of the world's largest quantitative research competitions, bringing together 156,000+ participants globally.

Participants have the opportunity to develop quantitative research skills, challenge themselves alongside peers from around the world and connect with a global community of quantitative thinkers.

Build skills on WorldQuant BRAIN®