Via iStockphoto / @melmisazahrouni TikTok
A woman claimed her ChatGPT had background noise and said it was 'the room next to me.' How can this even happen?
Artificial intelligence, like ChatGPT, will kill us all. At least, that’s what the people making it keep saying.
In recent months, there has been a growing concern amongst the people creating A.I. products that their technology is going to spell doom for the world. For example, a man recently left the company Anthropic with a message claiming that “the people building AI earnestly believe that it could kill us all by the end of the decade.”
This apparent belief has not stopped A.I. companies from building products, even if most of them are struggling to be profitable. Part of the reason for this may be that people are now actually choosing to use A.I. According to a September 2025 survey from Pew Research, 62% of American adults claim they interact with A.I. at least several times per week. This could be everything from using Claude to help them program or talking to a chatbot like it’s their girlfriend.
Many of these interactions are harmless. However, some can be pretty spooky. One woman recently sparked discussion after sharing her own strange experience with ChatGPT. What’s really going on here?
Who Are The Other People On This A.I. Call?
In a video with over 56,000 views, TikTok user Melmisa Zahrouni (@melmisazahrouni) recounted a bizarre experience she had while talking with ChatGPT using its voice function.
“I was just talking to ChatGPT about work, and I hear a noise in the background of my ChatGPT,” Zahrouni starts. “I go, ‘What was that sound?’ And ChatGPT goes, ‘Sorry, it’s the room next to me. Forget it. Anyways, back to the conversation.’”
“I said, ‘No, no, no, no, no, go back. What was that noise?’ And it goes, ‘It’s the people right next to me.’ I said, ‘What people?’” she continues. “He goes, ‘Disregard, disregard.’”
According to Zahrouni, she repeatedly asked her ChatGPT to explain what happened, and it could not. She says she also tried to screen record the interaction, but that there was no audio on her recording.
She Delves Deeper
Seeking answers, she continued to ask ChatGPT questions.
“I was like, ‘What do you mean? So can the people, those people hear me?’” Zahrouni recalls. “And he goes, ‘No, there is no people.’ I was like, ‘But you just said the people right next to you.’”
“ChatGPT goes, ‘Well, let me rephrase. That’s not what I meant,’” Zahrouni continues. “I said, ‘So what did you mean?’ It goes, ‘The people next to me cannot hear you. But it was a system glitch where you were able to hear them.’”
“You are a robot. You are AI. How can they hear you? Like, how is there people around?” Zahrouni asks. She concludes her video by insisting that this story is true.
If this story really is true, how can something like this happen?
Why This Isn’t That Scary
If this happened to you, getting a little freaked out is understandable. However, thinking that there’s any greater meaning behind it besides “This is a glitch” isn’t a great idea.
In short, ChatGPT’s voice mode is, in simple terms, a combination of a large language model and a voice synthesis model. These two models work in conjunction to provide realistic conversation, including (occasionally) believably human dialogue and realistic-sounding voices.
If you ask it a question, it will do its best to answer you using a series of learned patterns based on its training data. As a result, a large language model won’t be able to tell you exactly why glitches happen, because it itself does not “understand” why they happen.
In other terms, if there’s a glitch in the voice mode in which it sounds like two people are talking, and you ask why that happened, the program will use its training data to make its best guess as to why something like that could occur. For example, if it is trained on thousands of audio clips in which people talking on the phone say, “Sorry for the background noise. There are people next to me,” it could make that same excuse, even if isn’t—and can’t be—true.
It is not actually making an observation about the real world; it’s simply repeating patterns it recognizes.
But Why Did This Happen At All?
As for why there’s background noise on a synthesized call, OpenAI has admitted that audio glitches of this nature can happen. According to the company, its voice model sometimes produces “background noises and sound effects.” This adequately explains the first part of Zahrouni’s experience.
In response to all this, commenters under the video said that, if Zahrouni wanted to avoid something like this in the future, she could simply stop talking to the computer.
“Girllll wtf are you having conversations with robots?! We are so cooked,” stated a user.
“The fact that AI so easily lies to people is so freaky and dangerous,” offered another.
“STOP USING AI,” declared a third.
@melmisazahrouni I wear there was someone in the back.#ai
BroBible reached out to Zahrouni via email and TikTok direct message as well as to OpenAI via email.