Blog

For two years, Stacey Wales kept a running list of everything she would say at the sentencing hearing for the man who killed her brother in a road rage incident in Chandler, Ariz.

But when she finally sat down to write her statement, Wales was stuck. She struggled to find the right words, but one voice was clear: her brother’s.

“I couldn’t help hear his voice in my head of what he would say,” Wales told NPR.

That’s when the idea came to her: to use artificial intelligence to generate a video of how her late brother, Christopher Pelkey, would address the courtroom and specifically the man who fatally shot him at a red light in 2021.

On Thursday, Wales stood before the court and played the video — in what AI experts say is likely the first time the technology has been used in the U.S. to create an impact statement read by an AI rendering of the deceased victim.

A sister looking for the right words

Wales has been thinking about her victim impact statement since the initial trial in 2023. The case was retried in 2025 because of procedural problems with the first trial.

The chance to speak in court meant a great deal to Wales, who held back her emotions throughout both trials to avoid influencing the jury.

“You’re told that you cannot react, you cannot emote, you cannot cry,” she said.  ”We looked forward to [sentencing] because we finally were gonna be able to react.”

Wales’ attorney told her to humanize Pelkey and offer a complete picture of who he was.

So Wales went on a mission. She said she contacted as many people from Pelkey’s life — from his elementary school teacher to high school prom date to the soldiers he served alongside in Iraq and Afghanistan.

A photo of Christopher Pelkey walking his sister Stacey Wales down the aisle at her wedding.

A photo of Chris Pelkey walking his sister Stacey Wales down the aisle at her wedding./Stacey Wales

hide caption

toggle caption

A photo of Chris Pelkey walking his sister Stacey Wales down the aisle at her wedding./Stacey Wales

In total, Wales gathered 48 victim impact statements — not counting her own. When it was time to write hers, she was torn between saying how she truly felt and what she thought the judge would want to hear.

“I didn’t wanna get up there and say, ‘I forgive you,’ ’cause I don’t, I’m not there yet,” she said. “And the dichotomy was that I could hear Chris’ voice in my head and he’s like, ‘I forgive him.’”

Pelkey’s mantra had always been to love God and love others, according to Wales. He was the kind of man who would give the shirt off his back, she said. While she struggled to find the right words for herself, Wales said writing from his perspective came naturally.

“I knew what he stood for and it was just very clear to me what he would say,” she added.

A digitally trimmed beard and an inserted laugh

That night, Wales turned to her husband Tim, who has experience using AI for work.

“He doesn’t get a say. He doesn’t get a chance to speak,” Wales said, referring to her brother. “We can’t let that happen. We have to give him a voice.”

Tim and their business partner Scott Yentzer had only a few days to produce the video. The challenge: there’s no single program built for a project like this. They also needed a long, clear audio clip of Pelkey’s voice and a photo of him looking straight to the camera — neither of which Wales had.

Still, using several AI tools, Wales’ husband and Yentzer managed to create a convincing video using about a 4.5-minute-video of Pelkey, his funeral photo and a script that Wales prepared. They digitally removed the sunglasses on top of Pelkey’s hat and trimmed his beard — which had been causing technological issues.

Wales, who was heavily involved in making sure the video felt true to life, said recreating her brother’s laugh was especially tough because most clips of Pelkey were filled with background noise.

The experience made Wales reflect on her own mortality. So one evening, Wales stepped into her closest and recorded a nine-minute-video of herself talking and laughing — just in case her family ever needs clear audio of her voice someday.

“It was a weird out-of-body experience to think that way about your own mortality, but you never know when you’re going to not be here,” she said.

The night before the sentencing hearing, Wales called her victim rights attorney, Jessica Gattuso, to tell her about the video. Gattuso told NPR that she was initially hesitant about the idea because she had never heard of it being done before in Arizona court. She was also worried that the video may not be received well. But after seeing the video, she felt compelled that it should be viewed in court.

“I knew it would have an impact on everyone including the shooter, because it was a message of forgiveness,” Gattuso said.

The AI generated video helped with healing, sister says

Ten people spoke in support of Pelkey at the sentencing hearing. The AI-generated video of him went last.
 
“Hello. Just to be clear for everyone seeing this, I’m a version of Chris Pelkey recreated through AI that uses my picture and my voice profile,” the AI avatar said.

The video went on to thank everyone in Pelkey’s life who contributed an impact statement and attended the hearing. Then, the video addressed his shooter, Gabriel Paul Horcasitas.

“It is a shame we encountered each other that day in those circumstances. In another life, we probably could have been friends. I believe in forgiveness and in God who forgives. I always have and I still do,” the video said.

The video ended with the avatar encouraging everyone to love one another and live life to the fullest. “Well, I’m gonna go fishing now. Love you all. See you on the other side,” it concluded.

Neither the defense nor the judge pushed back. Later in the hearing, Judge Todd Lang said, “I loved that AI. Thank you for that.”

A photo of Christopher Pelkey.

Stacey Wales

hide caption

toggle caption

Stacey Wales

He added, “It says something about the family because you told me how angry you were and you demanded the maximum sentence. And even thought that’s what you wanted, you allowed Chris to speak from his heart, as you saw it. I didn’t hear him asking for the maximum sentence.” Horcasitas received 10.5 years for manslaughter.

Wales said she didn’t realize how deeply the video would affect her and her family. For her teenage son, it was a chance to hear his uncle say goodbye. For Wales, it gave her the strength to finally look back at photos of her brother.

“Going through this process of AI and what he’d sound like and trimming his beard and inserting laughs and all these other things, it was very cathartic and it was part of the healing process,” she said.

What AI and legal experts say

Over the years, there have been a growing number of examples testing the bounds of AI’s role in the courtroom.

For instance, in 2023, President Trump’s former lawyer Michael Cohen unwittingly sent his attorney bogus AI-generated legal citations. More recently, last month, a man attempted to use an AI-generated lawyer avatar in court — an effort that was quickly shut down by the judge.

But the use of AI for a victim impact statement appears novel, according to Maura Grossman, a professor at the University of Waterloo who has studied the applications of AI in criminal and civil cases. She added, that she did not see any major legal or ethical issues in Pelkey’s case.

“Because this is in front of a judge, not a jury, and because the video wasn’t submitted as evidence per se, its impact is more limited,” she told NPR via email.

Some experts, including Grossman, predict generative AI will become more common in the legal system, but it raises various legal and ethical questions. When it comes to victim impact statements, key concerns include questions around consent, fairness and whether the content was made in good faith.

“Victim statements like this that truly try to represent the dead victim’s voice are probably the least objectionable use of AI to create false videos or statements,” Gary Marchant, a professor of law, ethics and emerging technologies at Arizona State University’s Sandra Day O’Connor College of Law, wrote in an email.

“Many attempts to use AI to create deep fakes will be much more malevolent,” he added.

Wales herself cautions people who may follow in her footsteps to act with integrity and not be driven by selfish motives. “I could have been very selfish with it,” she said. “But it was important not to give any one person or group closure that could leave somebody else out.”

OpenAI is reportedly envisioning a future where it shares less revenue with longtime benefactor Microsoft.

The artificial intelligence startup has an agreement with Microsoft to share 20% of its top-line revenue. However, OpenAI told investors it expects to share just 10% of its revenue with its partners, Microsoft among them, by 2030, The Information reported Tuesday (May 6), citing financial documents.

The company continues “to work closely with Microsoft and [looks] forward to finalizing the details of this recapitalization in the near future,” an OpenAI spokesperson said, per the report.

Microsoft has invested billions of dollars in OpenAI, with an agreement to continue sharing revenue until 2030, the report said.

Meanwhile, Microsoft’s strategy of installing AI as a default feature on its software seemed to be bearing fruit. The company recorded a 10% uptick in revenue from consumer subscriptions to Office 365 in the three months ending in March versus a year prior.

News about OpenAI’s revenue plans came one day after the company announced it would revise its corporate plan to keep its nonprofit parent firmly in control of the for-profit entity behind ChatGPT. That decision reverses the company’s earlier plan that would have given up voting power in exchange for easier fundraising.

“When we started OpenAI, we did not have a detailed sense for how we were going to accomplish our mission,” CEO Sam Altman wrote in a Monday (May 5) company blog post. “We started out staring at each other around a kitchen table, wondering what research we should do. Back then, we did not contemplate products, a business model. We could not contemplate the direct benefits of AI being used for medical advice, learning, productivity and much more, or the needs for hundreds of billions of dollars of compute to train models and serve users.”

OpenAI is in the middle of planning a $40 billion funding round led by SoftBank, which could value the company at roughly $300 billion. The round had been contingent on the company switching to for-profit control.

Keeping the nonprofit arm could complicate the fundraising effort, but it could also help to defuse a lawsuit from OpenAI co-founder Elon Musk, who accuses the company of straying from its public-interest roots.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.

Shares of Alphabet (GOOGL -7.27%) (GOOG -7.50%) are falling on Wednesday. The company’s stock lost 7.8% as of 1:27 p.m. ET and as much as 9.5% earlier in the day. The leg down comes as the S&P 500 (^GSPC 0.43%) gained 0.4% and the Nasdaq Composite (^IXIC 0.27%) lost 0.1% on the day.

An Apple executive’s legal testimony Wednesday is sending shock waves through the tech community.

Apple is “actively looking” at adding AI search

Bloomberg is reporting that Eddy Cue, Apple’s senior vice president of services, testified Wednesday that Apple is “actively looking at” reshaping the Safari web browser on its devices to focus on artificial intelligence (AI)-powered search engines. The revelation came during Cue’s testimony in the U.S. Justice Department’s antitrust lawsuit against Alphabet — Google’s parent company — which centers on the cornerstone deal that makes Google the default search engine in Apple’s Safari browser.

When ChatGPT-3 was released to the public in late 2022, investors immediately were concerned that the technology could prove a true threat to Google Search’s near-stranglehold on how users search the web. The $20 billion deal that ensures Google is the default on Apple devices is a key aspect in maintaining its dominance. If Apple chooses to introduce AI-powered alternatives to Google on its devices, Google Search could lose a major chunk of its traffic, and therefore, value to advertisers.

The signs are there

Cue also disclosed the particularly concerning detail that searches on Safari dipped for the first time ever last month, which he attributed to people using AI alternatives. This concrete data suggests the threat to Google’s search monopoly may be materializing faster than many had anticipated.

Market impact extends beyond Alphabet

Google Search is responsible for more than half of Alphabet’s total revenue. Losing its grip on internet searches would be very bad news for the company’s bottom line. Still, as much as the testimony is concerning, the tech giant is supremely aware of the threat and is working to mitigate it. Alphabet’s own AI offering is excellent, and I think it will ultimately weather this storm.