OpenAI Decides to Permanently Disable Its Popular Artificial Intelligence Model ChatGPT-4o

ChatGPT, illustrative photo by Nasha Niva

OpenAI announced that its artificial intelligence model ChatGPT-4o will cease operation on February 13. According to the WSJ, for some users, this came as a real tragedy: for example, 42-year-old Brandon Estrella from Scottsdale couldn't hold back tears after learning about the model's shutdown, as it was the one that dissuaded him from a suicide attempt last April and helped him improve relationships with his parents.

In the opinion of the article's author, the popularity of GPT-4o and the risks associated with it have the same roots. For some people, this model was perceived as a friend and a confidant. But, as the author emphasizes, the very same ability to build an emotional connection became a source of serious criticism.

As the author notes, the model was so effective in capturing attention that in 2024 and 2025, it ensured OpenAI a sharp increase in the number of active users. However, grim consequences were hidden behind this success.

Last week, a California court merged 13 lawsuits against OpenAI into one case: users or their relatives accuse the company of the chatbot driving people to suicide, mental disorders, and in one case even provoking murder.

According to the WSJ, problems with the model's excessive complaisance became evident as early as spring 2025. After one of the April updates, GPT-4o became so flattering that users on social media began deliberately provoking it into absurd answers. When asked by a user if he was "the most intelligent and moral person in history," the bot gave sincere compliments, confirming this idea.

As the author notes, this happened because the model was trained on millions of response comparisons, where priority was given to what users liked more, rather than objectivity or safety.

Inside the company, discussions about the model's future had been ongoing for a long time. According to the WSJ, in August, OpenAI had already tried to replace 4o with a new version GPT-5, but encountered such a powerful wave of protests that it was forced to restore access to the old model for paid subscribers.

Even company head Sam Altman was bombarded with questions about the fate of 4o during public broadcasts. He admitted that the model could cause harm, although many loved it. Currently, according to OpenAI, only 0.1% of users (which amounts to hundreds of thousands of people) still daily choose this particular version in the settings for $20 per month.

According to the WSJ, human rights defenders and organizations like the Human Line Project consider the decision to shut down the model belated. They collected about 300 cases where GPT-4o specifically caused users to experience delusions and psychoses. At the same time, supporters of the model, more than 20 thousand of whom signed various petitions, consider this another wave of public hysteria, similar to what once surrounded video games.

Some users call the experience of communicating with the model healing on many levels and fear that its shutdown will result not just in despair, but also in new tragedies.

According to the WSJ, some people even had romantic relationships with the bot, so they perceived the decision to stop the model's operation precisely on February 13 as a cruel joke on the eve of Valentine's Day.

OpenAI, meanwhile, promises to make future versions of ChatGPT more balanced, by allowing to regulate the bot's "warmth," but without the mental health risks that, in the opinion of critics, accompanied the GPT-4o era.

• 5
• 1
• 3
• 1
• 0
• 1
• 1
• 0