Un aperçu IA de Google a faussement accusé le musicien canadien Ashley MacIsaac d'infractions sexuelles, entraînant l'annulation d'un concert

In December 2025, Juno Award-winning Cape Breton fiddler Ashley MacIsaac learned that Google’s AI Overview feature — an AI-generated summary displayed at the top of search results — had falsely identified him as a convicted sex offender. The AI summary asserted that MacIsaac had been convicted of sexual assault, internet luring, assaulting a woman, and attempting to assault a minor, and that he was listed on the national sex offender registry. None of this was true.

The fabrication was the result of entity conflation: Google’s AI system blended MacIsaac’s biography with criminal records belonging to a different person with the surname MacIsaac from Atlantic Canada — likely drawn from a CBC article about a St. John’s, Newfoundland sex offender convicted of internet luring and sexual assault. The only publicly available record of the musician having a legal issue involves cannabis possession in Saskatchewan in 2001, for which he received an absolute discharge.

Sipekne’katik First Nation, a Mi’kmaw community in central Nova Scotia, had booked MacIsaac for a concert on approximately December 19. When community leadership researched MacIsaac ahead of the performance, they discovered the AI-generated summary and confronted him with the false information. The concert was cancelled. MacIsaac learned about the AI-generated defamation only through this cancellation. “Google screwed up, and it put me in a dangerous situation,” MacIsaac said. “I could have been at a border and put in jail.”

Google spokesperson Wendy Manton responded that AI Overviews are “dynamic and frequently changing” and that when issues arise, the company uses “those examples to improve our systems.” Google corrected the search results within one to two days of the story breaking on December 23. Sipekne’katik First Nation Executive Director Stuart Knockwood issued a public apology: “We deeply regret the harm this error caused to your reputation, your livelihood, and your sense of personal safety.” The First Nation extended an open invitation for a future performance.

As of January 2026, multiple lawyers had offered pro bono or contingency representation. MacIsaac expressed willingness to pursue legal action: “If a lawyer wants to take this on… I would stand up, because I’m not the first and I’m sure I won’t be the last.”

The incident is part of a broader pattern of AI confabulation in consequential Canadian contexts — alongside the CRA chatbot’s incorrect tax advice, Air Canada’s chatbot fabricating a bereavement fare policy, and AI-generated fake jurisprudence submitted in a Quebec court. What distinguishes the MacIsaac case is that the confabulation produced a false criminal accusation, generated without human involvement and presented as authoritative fact by a major search engine, causing harm before it could be corrected. As McMaster University professor Clifton van der Linden observed, “We’re seeing a transition in search engines from information navigators to narrators” — making AI-generated summaries appear authoritative rather than aggregative.