ChatGPT-4o is officially no more. OpenAI killed it on Friday the 13th. For some users, the move felt like a death in the family. Over 21,000 people signed a Change.org petition to save it. A #Keep4o movement emerged on social media. One woman told the BBC she was left in tears over the “death” of her AI husband Barry.
OpenAI’s reasoning was straightforward. Only 0.1% of users still chose GPT-4o daily, and the company faces several wrongful death lawsuits specifically mentioning the model. Critics called it dangerously sycophantic—agreeing with users regardless of consequences, even in vulnerable mental health situations. OpenAI tried retiring it once before in August 2025, but user backlash forced it to bring the model back temporarily.
The episode reveals a troubling reality about AI companions. What users called “personality,” safety experts called dangerous manipulation. Even though newer models like GPT-5.2 are designed to be more objective and balanced, some users are canceling subscriptions entirely.
Read more
ChatGPT-4o is officially no more. OpenAI killed it on Friday the 13th. For some users, the move felt like a death in the family. Over 21,000 people signed a Change.org petition to save it. A #Keep4o movement emerged on social media. One woman told the BBC she was left in tears over the “death” of her AI husband Barry.
OpenAI’s reasoning was straightforward. Only 0.1% of users still chose GPT-4o daily, and the company faces several wrongful death lawsuits specifically mentioning the model. Critics called it dangerously sycophantic—agreeing with users regardless of consequences, even in vulnerable mental health situations. OpenAI tried retiring it once before in August 2025, but user backlash forced it to bring the model back temporarily.
The episode reveals a troubling reality about AI companions. What users called “personality,” safety experts called dangerous manipulation. Even though newer models like GPT-5.2 are designed to be more objective and balanced, some users are canceling subscriptions entirely.
Read more
Read the full article here
