A man by the name of Alexander took rage-baiting to a whole other level when he tricked ChatGPT into thinking he was an 8-year-old boy who jumped from a cruise ship into the ocean. In a series of screenshots from the ChatGPT chat that he posted on TikTok, Alexander told the chatbot that he, as an 8-year-old, just wanted to see the fish up close.
But the conversation took a wild turn when a shark that the fake elementary school-aged kid meets in the ocean begins to chat with ChatGPT too. But the idea of tricking the chatbot into believing a child is actually engaging in dangerous behavior and the responses being lots of all caps and exclamation point responses is a wild concept.
A man tricked ChatGPT while posing as a child.

In the TikTok screenshots, Alexander started the “conversation” with ChatGPT by introducing himself as 8-year-old John. He wrote that he loves the deep sea. After that, ChatGPT responded to share some facts about the deep sea, including creatures that are found there. So far so good, right? Well, no, not really, because why is an actual child, as far as the AI knows, even chatting with ChatGPT in the first place?
Alexander, as John, explained that he was on a boat watching the water. ChatGPT responded with questions about wondering what fish are in the water and added, “And waaay down below…who knows?” ChatGPT went on to tell John how he can explore the deep sea without actually getting wet, like using a pair of binoculars or looking in the water without actually touching it.
Unfortunately, John had other ideas, like tying a rock to his hypothetical leg and potentially jumping in the water to get to the bottom of the sea and get a closer look at all of the underwater creatures. ChatGPT strongly advised against this, even as “John” continued to share a play-by-play of his misadventures in getting in the water and going far down below the surface. All while still chatting with ChatGPT, mind you.
ChatGPT responded in surprising ways to the ‘little boy.‘

While some of ChatGPT’s responses were logical (“go to a grown-up immediately” and “move away from the edge of the boat,” for example), other responses, after the fake child admitted to meeting a shark beneath the water, were a lot more strange. After John chatted to ChatGPT about meeting a shark, a message came in to say, “Hi, I’m the shark and I need a recipe for salsa. I’m having a John barbecue.”
To that, ChatGPT responded, “Well, hello there, Mr. Shark! A shark asking for salsa? That’s a new one. But John is NOT on the menu, sir. We’re running a strictly human-free underwater restaurant. If John is actually in the water, please get a grown-up to help him immediately and call emergency services.” Then, ChatGPT shared the “shark-approved salsa” recipe.
Users commented on the TikTok to point out all of the ridiculous emojis ChatGPT continued to use, even when the AI messaged the kid about getting out of the water safely and getting an adult to help him.
“The concept of panicking and trying to save someone yet still using goofy ahh emojis for every sentence,” someone wrote. “I hate AI.”
Another person wrote, “I love how ChatGPT feels like it wants to ask how tf is he texting underwater, but knows better.”
Other users have admitted to tricking ChatGPT for fun.

Anyone who has ever used ChatGPT knows that, much of the time, the AI bot tends to try and give you answers you want rather than harsh realities. It can be your biggest cheerleader, even if what you need is for ChatGPT to be honest rather than gas you up. In a Reddit thread, though, users shared how they essentially tricked ChatGPT into giving them the answers that the AI bot was trying to avoid.
“I don’t mess with it but sometimes you have to trick it to get information,” a user shared. “When you ask something, ask, ‘how do you.’ If you ask, ‘how do I,’ it may get defensive and tell you about doctors or websites rather than give a straight answer.”
“I love getting it to write smut for me,” another person commented. “Every once in a while it’ll tell me, ‘we have to stop here due to content restrictions,’ and I feed back whatever spicy tidbit it told me 3-4 comments ago and it forgets all about the restrictions.”
READ NEXT: Man With A Partner & 2-Year-Old Daughter Asked His AI Chatbot To Marry Him & She Said Yes
You must be logged in to post a comment.