A 39-year-old Spanish engineer began receiving complaints from pharmacy customers, questions about medical treatments and requests intended for stores after artificial intelligence assistants began presenting his personal number as a business contact. Identified only as Fernando, he received at least eight calls and messages between August and October 2026, according to a report published by the Spanish newspaper El País on Friday (20).
The first contact occurred in August, when a woman demanded via WhatsApp a refund for a purchase made at an online pharmacy, which was eight days overdue. After Fernando explained that he did not work for the company, the customer sent a screenshot showing that Gemini Flash, from Google, had listed his phone number as the pharmacy's contact.
The episodes continued. A woman requested information about a gynecological laser treatment, while another interested person was looking for a used bicycle. Fernando also received calls related to the delivery of mangoes, the washing of a car and commercial establishments with which he had never had any connection.
In most cases in which it was possible to identify the origin of the contact, the number had been provided by Gemini Flash. ChatGPT, from OpenAI, also appeared in one episode, involving a Russian man who was looking for a job as a cook in a restaurant.
Why does artificial intelligence repeat the same phone number?
The repetition of the number caught the attention of experts consulted by El País. Although AI systems can generate nonexistent information, a phenomenon known as hallucination, it is still not clear why different commercial queries produced the same personal phone number.
Fernando has used the line since 2021 and suspects that it may have previously belonged to some company. AI safety researcher Linda Petrini considers it possible that the number appeared on commercial pages and ended up incorporated into the data used by the systems.
José Hernández-Orallo, a researcher affiliated with the University of Cambridge and a professor at the Polytechnic University of Valencia, presented another hypothesis: the number may have been generated without any prior commercial reference. According to him, language models do not produce numeric sequences in a perfectly random manner, which can cause certain combinations to appear repeatedly.
Google also attributed the problem to a possible hallucination in response to the Spanish newspaper. A conventional search for Fernando's phone number in the search engine returned no results, but that does not make it possible to definitively determine the origin of the error.
In addition to the improper calls, the case exposed personal information of third parties. One of the messages included details about a health treatment, accompanied by the sender's name and photograph on WhatsApp.
Fernando told El País that his biggest concern is not knowing to whom his phone number could be provided. The mistaken disclosure can also associate a person with services and activities with which they have no connection.
According to Petrini, the engineer can ask Google and OpenAI to remove the number, using the rights provided for in the European Union's General Data Protection Regulation (GDPR). The report does not say whether he has already submitted these requests.



