An influencer learned the limits of AI the hard way.

In a video posted to TikTok on Aug. 13, Spanish content creator Mery Caldass broke down crying in the airport after missing her flight because she did not have the necessary paperwork prepared to travel for a romantic getaway with her partner. And the reason, she claims, is that she looked to ChatGPT for advice.

“I asked ChatGPT and he said no,” says Caldass, in Spanish, about whether she would need a visa to make the journey to Puerto Rico.

(While a visa is not required for European Union citizens to enter Puerto Rico if you are staying for fewer than 90 days, travelers still must complete the Electronic System for Travel Authorization, an online application that determines the eligibility of visitors to enter the United States, per the State Department.)

“That’s what I get for not getting more information,” the influencer said through tears.

“I don’t trust that one anymore,” she added, referring to her AI assistant.

In the comments of her video, which has racked up thousands of views, many were not very sympathetic to the influencer’s situation.

“government websites exist to give you this information, if you can’t take the initiative to search those yourself then you are a lost cause atp,” wrote one.

“Why didn’t you use Google?” chimed in another in Spanish.

“Don’t you realize that Puerto Rico is like the U.S. when it comes to visas?” another said in Spanish.

But it appears that Caldass and her partner’s travel plans weren’t too disrupted by the mishap. The following day, the influencer shared a video from a Bad Bunny concert in Puerto Rico. And later in the week, she and her partner continued to update her followers about her travels throughout the territory.

Last month, OpenAI, the company behind ChatGPT, claimed that the chatbot now “thinks and acts” after a new update. The tool, which the company claimed was used by 500 million people every week, is designed to “intelligently navigate websites” and “conduct analysis,” the company said in a statement.

As the platform has exploded in popularity over the last two years, it has boasted several incredible achievements. Last year, for instance, a woman said that ChatGPT diagnosed her blood cancer a year before her doctors. Another young boy had been struggling with chronic pain for three years, and in 2023, ChatGPT identified his illness from his symptoms when his doctors were unable.

But ChatGPT has also been known to make mistakes, sometimes referred to as “hallucinations.”

Earlier this year, for example, a man who consulted ChatGPT about his diet found himself on involuntary psychiatric hold after he poisoned himself by substituting sodium chloride, or table salt, for sodium bromide, following the advice of his AI chatbot.

In a case report for the Annals of Internal Medicine: Clinical Cases, researchers cautioned that AI can “potentially contribute to the development of preventable adverse health outcomes.” (PEOPLE)