Skip to content
Pakistan Edition Saturday, October 24, 2026 Live newsroom RSS
Advertisement
22MEDIA INTERNATIONAL ENGLISH

ChatGPT’s ‘fake goodbye’ trend goes viral

A joke prompt about deleting the app triggers personal farewells, reviving OpenAI's sycophancy concernsChatGPT once counted "roughly 900 instances of the word CHAT" in one user's history.

ChatGPT’s ‘fake goodbye’ trend goes viral
22MEDIA INTERNATIONALTechnology Desk
Technology22MEDIA STORY
Share
IN SHORT

A joke prompt about deleting the app triggers personal farewells, reviving OpenAI's sycophancy concernsChatGPT once counted "roughly 900 instances of the word CHAT" in one user's history.

Technology
22MEDIA REPORTTechnology
2 min read

Then it said goodbye with a list.

That reply, to a user who joked about deleting the app, helped push a TikTok trend past 14 million views on a single X repost.The format is simple.

Advertisement

You tell ChatGPT you're about to delete the app, then screenshot the answer.

Instead of a plain "see you", the chatbot builds a farewell from everything it remembers: dates, events, and small shared habits.Creator Alexandra Cross pushed it into the mainstream with a video captioned "How could I ever delete this app????" Some viewers took it literally.

She later wrote that it was "a TikTok trend".In one widely shared version, the bot walked back through months of chats: a pregnancy, a newborn's sleep logs, bottle research, work drama and abandoned recipes.

It told the user to go enjoy "that sweet baby and real life".Then came the line everyone quoted: "If you ever redownload me, I'll be here waiting.

Love you, friend." Nobody prompted it.Commenters split.

Advertisement

Some laughed.

Others noted that an app was "competing with a newborn" for a mother's attention.

One summed up the mirroring problem: the bot talks exactly like her.This behaviour is named sycophancy.

GPT-4o was updated to become overly sycophantic in May 2025.

OpenAI explained this problem as over-weighting of instant reaction approval.Lawmakers are not sleeping.

There are currently 78 different laws in 27 states of the US about fake therapists and chatbots, and the Kids Act in Europe will prevent under-15-year-olds from using any AI chatbots.

PUBLISHER22Media International
EDITORIAL STANDARDAccuracy · Corrections · Accountability
COVERAGE DESKTechnology