More than ten coaches carrying tourists arrived in Jurilovca, Romania, in September 2026 for the Lipovan Borscht Festival. Numerous other visitors travelled to the village in private cars. There was only one problem: the festival was not taking place.
According to Jurilovca Mayor Eugen Ion, tourism agencies had promoted trips to the event on social media without confirming the information with the local authority responsible for organising the festival. Some tourists also said they had asked ChatGPT whether the event was taking place and received an affirmative answer.
Representatives of the local administration subsequently tested ChatGPT themselves. The platform reportedly indicated that the festival would be held on September 4 and 5, despite the absence of an official announcement and the fact that the 2026 edition had not been organised.
The incident, reported by Romanian media, is a practical example of how inaccurate information can circulate between websites, social media, commercial offers and generative AI platforms until it appears credible enough to influence real-world decisions.
It also demonstrates why Generative Engine Optimisation is not only about convincing ChatGPT, Gemini, Perplexity or other AI platforms to mention a brand. GEO must also help generative engines identify which information is official, current and supported by authoritative sources.
ChatGPT Did Not Create the Entire Information Problem
Blaming ChatGPT alone would oversimplify what happened in Jurilovca.
According to the mayor, tourism agencies had already advertised trips to the festival. Their promotional materials mentioned different dates, including weekends in late August and September. The agencies reportedly did not contact Jurilovca Town Hall to establish whether the festival would take place.
The AI platform therefore entered an information environment that was already affected by unverified claims.
A generative engine researching the event could encounter pages about previous editions, recurring date patterns, tourism offers and social media posts presenting the festival as current. If official information about the cancellation was less visible or harder to interpret, the system could produce a plausible answer by combining outdated and inaccurate details.
This does not remove the platform’s responsibility for generating an unsupported response. It shows, however, that an AI hallucination may be the final stage of a longer information failure.
In the Jurilovca case, several failures appear to have occurred:
- Tourism agencies promoted an event without confirming it with the organiser.
- Different festival dates circulated on social media.
- Users treated AI-generated information as factual confirmation.
- ChatGPT reportedly supplied specific dates without a reliable official source.
- Commercial operators and travellers failed to complete a final verification.
This is precisely where AI SEO and GEO become relevant. Organisations must consider not only how traditional search engines display their information, but also how AI platforms retrieve, interpret and reproduce it.
Why Generative AI Can Confuse Past and Current Events
Recurring events are particularly difficult for generative engines to interpret.
A festival organised annually may have dozens or hundreds of online references containing its name, location and dates from previous editions. Old articles can continue to rank in traditional search results. Event directories may preserve expired listings without clearly identifying them as historical. Tourism websites may reuse previous descriptions or change the year without verifying the rest of the information.
Social media creates additional ambiguity because posts can be copied, republished and transformed into new promotional materials. Once an incorrect date appears on several pages, repetition can make it look independently confirmed even when every version originates from the same unsupported claim.
Generative AI systems do not think or verify facts as humans do. They produce answers based on patterns, retrieved information and the context available when a user submits a question. OpenAI itself acknowledges that language models can produce plausible but false statements, a phenomenon commonly described as an AI hallucination.
When current official information is missing, unclear or less prominent than outdated content, the risk of an incorrect answer increases.
GEO Is Also About Information Accuracy
Many businesses understand GEO primarily as a visibility strategy: the brand should be mentioned in AI-generated answers relevant to its products, services or industry.
Visibility is only part of the objective.
An organisation gains little from being visible if the information associated with its name is wrong. A hotel does not want AI platforms to provide an outdated address. A manufacturer does not want a discontinued product presented as available. A clinic cannot afford to have obsolete contact information reproduced in a generated answer. An event organiser does not want tourists directed to an event that is not taking place.
A complete GEO strategy should address three connected objectives:
Visibility
The organisation must be sufficiently relevant and authoritative to appear in generative answers.
Accuracy
The information associated with the organisation must be current, consistent and supported by primary sources.
Attribution
Generative engines should be able to identify and cite the organisation’s official content rather than relying on copied, outdated or unauthorised versions.
The best practices for Generative Engine Optimisation identified by TUYA Digital include comprehensive content, relevance, authority, clear structure and factual support. These elements help generative platforms understand both a page’s subject and the credibility of its information.
They cannot guarantee that an AI system will never produce an incorrect answer. They can, however, reduce ambiguity and strengthen the signals associated with the authoritative version of a fact.
Cancellation Announcements Must Be Optimised
Organisations usually invest significant resources in promoting an event but pay much less attention to communicating a cancellation, suspension or change of date.
A short social media post may be considered sufficient. The original event page may remain online without an update, or it may be deleted completely. Partner websites, ticket platforms and event directories may continue displaying the previous information.
From a GEO perspective, this creates a dangerous information gap.
A cancellation announcement should be treated as the definitive source for the event’s current status. It should be published on a permanent, indexable page and contain direct statements that users, search engines and generative platforms can interpret without additional context.
The page should clearly include:
- The complete name of the event
- The edition and year affected
- The original dates, where applicable
- The explicit status: cancelled, suspended or postponed
- The identity of the organiser
- The date of the official announcement
- Information about tickets or reservations
- Official contact details for verification
- The status of any future edition
A statement such as “the festival will not take place this year” may be clear at the moment of publication, but it becomes ambiguous when accessed later. A stronger formulation would state that “the 2026 edition of the Lipovan Borscht Festival in Jurilovca will not take place.”
This direct-answer format also supports Answer Engine Optimisation, which helps AI-powered systems identify and extract a concise response to a specific question.
Official Information Must Be More Visible Than the Rumour
Publishing accurate information does not automatically make it dominant.
A false event listing may be copied across multiple websites, attract social media engagement and receive links from tourism pages. Meanwhile, the official correction may exist only in a single social media post with a vague title.
For a user, the correction technically exists. For a generative engine processing several competing sources, the information environment may still appear contradictory.
This is why relevance, trust and credibility matter in GEO. The official source must be clearly identifiable, consistently updated and connected to the wider network of information about the event.
The original event page should remain accessible but display a prominent cancellation notice. That notice should link to the complete official announcement. The announcement should, in turn, identify the original event and the dates that are no longer valid.
Organisers should also contact tourism agencies, ticket platforms, media publications and event directories carrying the incorrect information. Correcting only the official website leaves the contradictory third-party content active.
GEO therefore overlaps with SEO, digital public relations, reputation management and crisis communication. The goal is not merely to publish the truth but to make the authoritative version visible and understandable across the entire information ecosystem.
Structured Data Must Be Updated
Event structured data can help search engines and AI-powered platforms interpret dates, locations, organisers and event status. It can also become another source of misinformation if it is left unchanged.
When an event is cancelled or postponed, its structured data should reflect the new status. The visible content and machine-readable information must communicate the same facts.
Deleting the entire event page is not always the correct solution. If the URL disappears, users and generative engines may lose the official source that can confirm the cancellation. Older third-party pages can then remain visible without an authoritative page contradicting them.
Keeping the page online with an explicit status usually provides a clearer signal than redirecting visitors to a general homepage or returning a 404 error.
Structured data alone will not solve the problem. If the visible page, metadata, social posts and partner listings contain conflicting information, technical markup cannot create consistency by itself.
GEO and AEO Must Work Together
The Jurilovca incident demonstrates the practical relationship between GEO and AEO.
AEO helps a page provide a clear and extractable response to a direct question:
“Is the 2026 Lipovan Borscht Festival taking place?”
“No. The 2026 edition of the Lipovan Borscht Festival in Jurilovca has been suspended and will not take place.”
GEO provides broader context that helps a generative engine evaluate and use that answer. This includes the authority of the source, the completeness of the information, supporting evidence, consistent entity details and references from other trusted websites.
A short answer without sufficient authority may be ignored. A comprehensive page without a clear answer may be misinterpreted. Using both approaches improves accessibility and credibility.
Organisations Must Monitor Their Presence in AI Answers
Companies already monitor search rankings, reviews, media mentions and social conversations. They must now monitor how generative platforms represent their brands, services, products and events.
Monitoring should include the questions users are most likely to ask:
- Is the event taking place?
- What are the confirmed dates?
- Where is it being organised?
- Has it been cancelled or postponed?
- Where can tickets be purchased?
- Who is the official organiser?
- Which source supports the answer?
Check answers across several platforms because results can vary depending on the model, retrieval system, prompt, location, and available sources.
Specialised GEO analytics tools can help organisations evaluate how frequently they appear in generative answers and whether their websites are cited. Manual testing remains necessary when information changes rapidly or can affect financial, medical, legal or travel decisions.
Monitoring is also essential after publishing a correction. Organisations should verify whether the updated information begins appearing in AI-generated answers and identify the sources still supporting the obsolete version.
GEO Cannot Guarantee That AI Will Never Be Wrong
No responsible GEO strategy can guarantee the complete elimination of AI hallucinations.
Generative platforms may rely on outdated data, fail to retrieve an official page, misinterpret a source or combine unrelated facts. Their models, retrieval systems and ranking mechanisms remain outside the control of individual organisations and GEO agencies.
GEO can improve the information environment in which these systems operate. It can make official information easier to discover, understand and cite. It can reduce contradictions, reinforce authoritative sources and help distinguish current facts from historical content.
The Jurilovca incident demonstrates what can happen when this information chain breaks. Unverified festival dates circulated through tourism advertisements and social media, an AI platform reportedly confirmed them, and tourists made real travel decisions based on an event that had never been officially announced.
The lesson for businesses and public organisations is clear: publishing accurate information is no longer enough. That information must also be structured, maintained, distributed and optimised for the generative systems increasingly used to make decisions.
TUYA Digital helps organisations understand how their brands and information appear in AI-generated answers and develops strategies to strengthen visibility, accuracy, and authority across generative platforms. Contact our SEO and GEO specialists to discuss how your organisation can prepare for AI-powered search.