Canadian fiddler Ashley MacIsaac says the ordeal left him worried for his safety.Rob Gurdebeke/The Globe and Mail
Cape Breton fiddler Ashley MacIsaac had a concert cancelled and is worried for his safety after he says Google incorrectly described him as a sex offender in an AI-generated summary last week.
He was arranging to play a concert last Friday at the Sipekne’katik First Nation a little north of Halifax when he learned that its leadership had changed their mind. They had read online, Mr. MacIsaac was told, that he had convictions related to internet luring and sexual assault.
That information is not true, and was later revealed to have been the result of Google’s AI-generated search summary blending MacIsaac’s biography with that of another man, who appears to be a Newfoundland and Labrador resident bearing the same last name. The Sipekne’katik First Nation has since apologized to MacIsaac, and Google has amended search results for the musician.
The error showcases the unintended consequences of AI-aggregated misinformation in an era when tech companies are racing to compete for dominance in the burgeoning sector. “Google screwed up, and it put me in a dangerous situation,” Mr. MacIsaac said in an interview.
The fiddler says the ordeal left him worried for his safety, theorizing that a victim of sexual assault could be triggered by the false information and confront him. He’s also concerned about work he may have lost without realizing it, if previous promoters and venues decided not to book him because of search-engine misinformation.
The professional consequences could extend to his ability to enter the U.S. for concerts, given increased social-media scrutiny from border agents. He sees his experience as a warning sign for others.
“People should be aware that they should check their online presence to see if someone else’s name comes in,” Mr. MacIsaac said.
The Juno Award-winning musician rose to fame in the nineties by fusing Celtic fiddle music with boundary-pushing elements of hip hop, electronic music and punk rock – while sometimes courting controversy for acting and living in ways that would be considered prosaic today, or for remarks he would later call sarcastic.
The only publicly available record of Mr. MacIsaac having a run-in with the law involves cannabis possession more than two decades ago, for which he was given a discharge.
The Sipekne’katik First Nation said in its apology to Mr. MacIsaac last Friday that he would be welcome to perform in their community in the future.
“We deeply regret the harm this error caused to your reputation, your livelihood, and your sense of personal safety,” wrote Stuart Knockwood, the First Nation’s executive director, on behalf of its chief and council. Representatives from Sipekne’katik First Nation did not respond to requests for comment.
The letter, which Mr. MacIsaac shared with The Globe and Mail, continued: “It is important to us to state clearly that this situation was the result of mistaken identity caused by an AI error, not a reflection of who you are.”
The information appearing on Google search results for Mr. MacIsaac’s name has been fluid since he posted about it on social media last week. Querying Mr. MacIsaac’s name alone no longer returns an AI-generated biography. On Thursday, a search query including both his name and a town connected to the other Mr. MacIsaac returned a result that said he had been “convicted of sexual assault.” That summary no longer appears.
In an e-mail, Google spokesperson Wendy Manton said: “Search, including AI Overviews is dynamic and frequently changing to show the most helpful information. When issues arise – like if our features misinterpret web content or miss some context – we use those examples to improve our systems, and may take action under our policies.”
The internet is under a vast restructuring as companies adopt generative AI into their products. Google, one of the world’s most valuable businesses, has been racing to include AI-generated search summaries as it seeks to maintain its lead, with services such as OpenAI’s ChatGPT nipping at its heels. (Canadian media organizations including The Globe and Mail sued OpenAI in 2024, for allegedly violating copyright law by scraping proprietary news content without consent or payment to train its models.)
Clifton van der Linden, an associate professor at McMaster University who has studied AI-generated misinformation, described Mr. MacIsaac’s situation as a consequence of a shift in public expectations of search engines in the AI age.
“We’re seeing a transition in search engines from information navigators to narrators,” Prof. van der Linden said. “I would argue that there’s evidence to suggest that AI-generated summaries are seen as authoritative by lay users.”
This, he says, can have pernicious consequences. But the current AI arms race, he says, does not incentivize accuracy; instead, it incentivizes companies such as Google to maintain their dominance.
“Google wants to replicate the strategy that led it to be the default search engine for the world – which is to produce a sufficiently reliable return to a sufficient segment of the population so as to become the default search engine that people go to to find information.”