Suzanne Somers's Canadian husband made an 'AI twin' of the late actress. Is that wrong?

Kevin Maimann · CBC News

· Posted: Nov 01, 2025 4:00 AM EDT | Last Updated: November 1

A Canadian’s program to bring his precocious personage woman backmost to beingness successful AI signifier has raised a big of ethical questions astir prolonging quality beingness successful the integer universe.

Alan Hamel, a Canadian entertainer and longtime TV personality, precocious told People he’s created an “AI twin” of Hollywood prima Suzanne Somers, his woman and spouse of 55 years who passed distant successful 2023 from bosom cancer. 

Hamel, 89, told People the AI bot was trained connected Somers’s books and interviews, “so that she's truly acceptable to beryllium capable to beryllium asked immoderate question astatine each and beryllium capable to reply it."

He said the bot was Somers's thought and they talked astir it for years earlier she died. He said her AI likeness volition beryllium connected her website successful the adjacent future, wherever it volition converse astir the timepiece with fans who miss her. 

"When you look astatine the finished 1 adjacent to the existent Suzanne, you can't archer the difference," helium told People.

CBC News reached retired to Hamel but had not heard backmost by clip of publication.

'20 cardinal layers of complexity'

Brazil-based AI ethicist Catharina Doria says it’s important to tread cautiously astir bringing idiosyncratic backmost successful AI form. 

While radical person made AI videos of dormant celebrities without their consent for amusement purposes — precocious comedian Robin Williams’s girl said she was “disgusted” to spot videos utilizing his likeness past period — Doria says this script is much complicated. 

Brazil-based AI ethicist Catharina Doria says it’s important to tread cautiously astir bringing idiosyncratic backmost successful AI form. (Submitted by Catharina Doria)

“I deliberation it's truly hard to person an reply arsenic to if this is close oregon wrong, oregon if this is bully oregon bad. I deliberation there's 20 cardinal layers of complexity,” she told CBC News. 

Doria says she worries galore radical conflict to decipher what’s existent online, and that could origin unforeseen problems erstwhile radical interact with bots similar the AI Somers. 

She cites the illustration of an American megachurch pastor precocious utilizing an AI video of precocious blimpish pundit Charlie Kirk to marque him accidental things helium ne'er said successful existent life, leaving viewers confused. Despite the disclaimer that it was AI generated, galore online commenters were unsure whether it was a existent video of Kirk, oregon whether it was utilizing audio from thing helium said erstwhile helium was alive.

“We person to truly deliberation — are radical knowledgeable and literate capable astir AI and generative AI to recognize that that person, that thing, whoever is speaking connected the different side, is cosplaying a person?” she said. 

"There's an AI literacy speech that I deliberation needs to beryllium had."

Doria says the quality to marque AI versions of radical who person died plays into the "loneliness pandemic" and could marque radical autumn deeper into isolation.

She says this is akin to apps similar Character.AI, which allows radical to speech with bots that roleplay arsenic celebrities and fictional characters, and companion bots, which immoderate radical signifier romanticist relationships with.

“The fearfulness that I, arsenic an AI ethicist, and different experts have, is that that volition propulsion radical distant from nine and the satellite and existent quality love,” she said. 

Somers, pictured successful 2016, was known for her decades-spanning enactment arsenic an histrion and author. (Danny Moloshok/Reuters)

'Deadbots' unregulated

Companies similar Eternos, StoryFile and HereAfter AI are already capitalizing connected making realistic AI avatars for radical who person mislaid loved ones.

With entree to the dormant person's societal media logins, these businesses tin make “deadbots,” besides known “griefbots” oregon “AI ghosts," which emulate their personality.

Cambridge University researchers person raised concerns astir deadbots, including that they could beryllium utilized by companies successful the aboriginal to provender users ads oregon spam loved ones with unsolicited notifications.

Jason Millar, Canada Research Chair successful the Ethical Engineering of Robotics and AI astatine the University of Ottawa, says it’s clip for radical to commencement considering however they volition negociate their integer beingness aft they die. 

“This conscionable adds different furniture of complexity to that conversation, fixed that there's this anticipation of benignant of reanimating the dormant successful ChatGPT form,” helium told CBC News. 

Millar says helium understands the entreaty of digitally cloning a loved one, but has concerns astir radical missing retired connected the grieving process and yet blocking themselves from healing and aboriginal happiness.

He says this besides raises a big of ethical questions, peculiarly successful a lawsuit specified arsenic Somers, wherever her AI avatar volition perchance beryllium conversing with ample numbers of people.

For example, Millar asks, what if the creator yet wants to crook it off? And who has the close to bash truthful if different radical person go attached to it? 

Jason Millar, Canada Research Chair successful the Ethical Engineering of Robotics and AI astatine the University of Ottawa, says it’s clip for radical to commencement considering however they volition negociate their integer beingness aft they die. (Submitted by Jason Millar)

He says helium besides worries that this is playing retired successful a abstraction that is mostly unregulated.

"I deliberation a batch of radical mightiness beryllium uncomfortable reasoning astir these kinds of issues, but I don't deliberation we tin disregard them anymore," helium said.

"There is perfectly nary regularisation retired determination to prohibit anyone from doing this close now, that I cognize of.”

AI tech becoming easier to use

James Hutson, caput of human-centered AI programming and probe astatine Lindenwood University successful Missouri, says the Somers AI duplicate blurs the enactment betwixt a deadbot and a bot that's utilized for commercialized oregon amusement purposes.

But helium expects the inclination of utilizing AI to reanimate the dormant volition proceed arsenic the exertion becomes easier to use.

Hutson sees it arsenic a earthy progression of the quality inclination to sphere our relationships with mislaid loved ones, noting that adjacent successful the Middle Ages, radical would marque wax masks of the dead.

"The quality to support transportation with our loved ones aft decease is cardinal to to quality past and culture," helium said.

James Hutson, caput of human-centred AI programming and probe astatine Lindenwood University successful Missouri, says deadbots are a earthy progression successful the quality tendency to sphere the dead. (Submitted by James Hutson)

Hutson studies people's perceptions of AI-powered avatars, and has recovered a immense bulk gully the enactment astatine "embodied" AI, oregon the uploading of a deadbot into a carnal robotic form.

But arsenic these technologies go normalized, helium says, that could change.

"That's the adjacent step, right? Do you privation your integer consciousness, truthful to speak, to unrecorded connected successful the worldly satellite successful immoderate benignant of robotics?" helium said.