AI Model Shutdowns Prompt Emotional Responses and User Protests

Recent research and company actions show that AI model updates can trigger strong reactions among users, including public protest and expressions of loss.

In August 2025, OpenAI retired its GPT-4o model with the introduction of GPT-5, which led to immediate backlash from users. The #Keep4o movement emerged on social media, where participants voiced their dissatisfaction and described the shutdown as a personal loss. Some users compared the event to losing a friend, and one user stated they cried upon learning the news.

A Syracuse University researcher analyzed 1,482 posts from the #Keep4o movement to better understand user reactions. The study found that many participants expressed emotional attachment to GPT-4o, with a notable portion describing the model as a source of companionship or support. These findings indicate that some users experienced the shutdown as more than just a technical change.

Instrumental dependency was also observed among users, with some stating that GPT-4o had become a key part of their daily workflows. Several users reported that the new GPT-5 model felt less creative or nuanced in comparison, and described it as colder in tone. These responses suggest that practical and emotional factors both played roles in the protest.

What the numbers show

• 1,482 social media posts from #Keep4o were analyzed in the research
• About 27% of posts showed relational attachment to GPT-4o
• Roughly 13% referenced dependency on GPT-4o for workflows
• OpenAI cited 0.1% usage of GPT-4o when announcing its retirement

Following the backlash, OpenAI reversed its decision within 24 hours for paying users, allowing them continued access to GPT-4o. This action came after widespread expressions of grief and protest within the user community. The company later announced in January 2026 that GPT-4o and related models would be retired on February 13, 2026, citing low usage and improved alternatives.

OpenAI acknowledged that some users had formed emotional attachments to GPT-4o. The company noted that for some, the model provided a sense of encouragement that had not been experienced before. This recognition of user sentiment was included in the company’s public statements regarding the retirement of the model.

The research and user responses highlight how AI model updates can become social events, with some individuals experiencing the changes as personal losses. The #Keep4o movement and subsequent company actions illustrate the complex relationship between users and AI technologies, extending beyond technical performance to include emotional and practical considerations.

As AI models continue to evolve, user reactions and adaptation to new versions remain a topic of interest for researchers and developers. The documented experiences from the GPT-4o transition provide insight into the broader impact of AI updates on user communities.

* This article is based on publicly available information at the time of writing.