In 2025, an AI-written travel article invented hot springs in Tasmania; visitors contacted and arrived at a local hotel asking about pools that did not exist

In July 2025, a travel website published what seemed to be a guide to one of Tasmania’s alluring experiences: “Weldborough Hot Springs,” described as a secluded forest retreat with pools that are rich in therapeutic minerals. However, there was one problem: the retreat did not exist. According to ABC News, the article appeared on Tasmania Tours, a booking website operated by Australian Tours and Cruises, and it remained online long enough for travellers to drive to the remote north-eastern Tasmanian town while looking for pools that did not exist in real life. This incident has since become an unusually concrete example of how AI-generated travel information can turn a fabricated destination into a believable place.

AI created a destination that did not exist in the real world

The fictional attraction appeared in a page titled "7 Best Hot Springs Tasmania Experiences for 2026." ABC News reported that the page described Weldborough Hot Springs as a “peaceful escape” and an “authentic connection to nature,” while claiming visitors could find pools containing healing minerals. Moreover, the website used AI-generated images depicting hot springs that did not exist at the locations. It is worth noting that while Weldborough is a real small community in north-eastern Tasmania, the Weld River running through it is cold and there are no hot springs. Therefore, the fabricated attraction was built around a place with a real name, making the misinformation appear considerably more plausible to someone unfamiliar with the region.

Following the webpage’s recommendations, visitors began arriving at the Weldborough Hotel asking how to reach the supposed pools. Publican Kristy Probert told the publication that she was receiving daily calls and in-person enquiries, including from people seeking accommodations. One instance involved a group of 24 drivers from mainland Australia who reportedly diverted their trip to Weldborough specifically to find the non-existent hot springs. Instead of the mineral-rich pools or a secluded bathing site, they found the ordinary cold Weld River and the local pub.

The website had more AI-generated errors

The hot springs article was not an isolated mistake, as ABC News noted that several other Tasmania Tours materials contained AI-generated content and images with geographical errors. Some images purported to depict Tasmanian locations but showed landscapes that did not match the real places. One image reportedly claimed to show a view from kunanyi/Mount Wellington over Hobart, but instead it resembled a Brazilian landscape; another purported to show Salamanca Market with incorrect features. According to the company’s owner, Scott Hennessy, some of the marketing materials had been outsourced to a third-party company using AI, and although posts were usually reviewed, some were published accidentally while he was overseas.

Hennessy told ABC News that the company had been using artificial intelligence because it lacked the resources to produce enough fresh content by itself. He acknowledged that the technology could sometimes produce good results but could also get things completely wrong. The AI-generated blog posts had been reportedly removed after the problems emerged. The incident underscores the key weakness of generative AI: a fluent description and a convincing image can create the appearance of authority even when the underlying claim has never been verified.

The Weldborough incident also fits with recent research examining AI hallucinations in travel. A 2026 study published in the Journal of Consumer Behaviour titled “AI Hallucinations in Tourism: How Errors Impact Consumer Trust and Recommendation Acceptance,” used two experiments to observe how fabricated information in chatbot-generated itineraries affects people’s willingness to follow recommendations. The researchers found that hallucinations can influence behaviour through perceptions of accuracy, usefulness and trustworthiness. The research suggests that people’s willingness to follow AI-generated travel advice depends heavily on how accurate, useful and trustworthy the recommendation appears, even when errors are present.