Replacing human support staff with automated systems can be pretty frustrating for customers, as highlighted in a recent Instagram post. Evan, a Seattle-based activist who has recently been fighting against Artificial Intelligence, detailed in his post an alleged incident between an Amazon customer and the company’s AI chatbot. After the customer asked to be connected to a “live person,” Evan claimed that the chatbot directed them to a phone sex line instead.
Evan started his post by calling the number allegedly given to the customer, only to hear, “Hi, baby. I’m so excited you’ve joined my party. Hundreds of girls are on our chat system right now, just waiting for you to connect. You can do it all here for as low as 25 cents a minute with absolutely no connection fee.” He told his viewers that his information came from a trustworthy source. Then he inserted a screenshot of the alleged chat history between the bot and the customer.
According to Evan, the incident has been escalated to the desk of CEO Andy Jassy. Evan noted that, “Executives at Amazon are freaking out because one of their chatbots told an irate customer to call this number for customer service.” He further suggested that while the leadership might be frustrated, they are in a difficult spot because there is no human to blame for the mistake. As Evan put it, “They have no one he can fire because, well, it was a chatbot.”
He claimed the mistake was a result of chatbots not being as intelligent as advertised
Evan argued that these systems are not built to handle the nuance of human interaction. He alleged, “These kinds of things are just going to keep happening because the reality is chatbots are just nowhere near smart enough to understand how to deal with human situations, and they keep f—ing hallucinating s—.” Now, it should be noted that this entire post is Evan’s perspective and claims. We haven’t been able to verify this situation.
Evan followed his criticism with an accusation. “Where did this chatbot pull a phone sex line from to give to somebody? I don’t know, but I’m sure it was programmed not to let people contact a human in customer service,” Evan added. “And so it was just trying to find some number it could give because chatbots are also programmed to be amenable and helpful.”
Irrespective of the veracity of his claims, Instagram flooded the comments. One user commented on the irony of the interaction, “Chatbot: This customer is not happy….PHONE SEX LINE IS THE CORRECT ANSWER.” Interestingly, a few months ago, a Redditor hit the help button on his Apple mental health app, only to be taken to a gambling website.
Others are pointing out the frustration inherent in these systems, with one user noting, “I work in Customer service. And people like real people, they are sick of chat bots.” Another echoed this sentiment by saying, “When are these execs going to realize the reason we’re contacting support is because the issue can’t be fixed by self-service?” For instance, a TikToker recently shared that despite an Uber driver going AWOL, the automated system still wanted to fine him.
Providing an incorrect number isn’t an isolated problem for Amazon, either. A report by Gadget Review highlights a broader, more dangerous trend where AI tools are surfacing private or incorrect information because they are trained on massive, uncontrolled datasets.
In one instance mentioned by Gadget Review, a user had their personal phone number turned into a public directory for businesses they had never heard of. It led to a constant stream of unwanted calls. Another software engineer found his personal WhatsApp being flooded with messages because an AI model incorrectly identified his number as a support line for a company called PayBox.
Per the outlet, security experts are also warning that bad actors are actively manipulating these systems. Companies like Aura Labs have found that scammers are intentionally poisoning the web. They put up fake customer service numbers on sites like Yelp and YouTube, knowing that AI models will eventually scrape that data and present it as fact to users. As Qi Deng of Aura Labs warned, “Attackers are quietly rewriting the web that AI systems read.”
The reality is that we are dealing with systems that are designed to be helpful. However, they often lack the basic common sense required to distinguish between a legitimate support number and a scam or an inappropriate service. Until these platforms become more reliable, the advice from tech experts is simple: skip the AI and go directly to official company websites.
Evan summarized the current situation in his post, stating, “Anyway, until next time, remember AI is not intelligent. Chatbots are not your f—ing friends.”
Published: Aug 1, 2026 08:53 pm