OpenAI has faced criticism from paying users over recent changes to ChatGPT. Many users say they are redirected to a more conservative AI model when emotionally or legally sensitive topics arise in chat.

OpenAI executive releases statement

Reddit threads on the subject claim that the new security measures implemented this month negatively impact the user experience. Users say ChatGPT switches models in certain situations, but these switches are not transparent and are completely beyond the user’s control.

One Reddit user summarized the situation as follows:
“Adults have the right to choose the model that suits their workflow, context, and risk tolerance. Instead, we are faced with silently overridden settings, hidden security prompts, and a model selector that has become a stage prop.”

Following the growing backlash, OpenAI executive Nick Turley issued a statement on social media. Turley stated that the security prompts are only activated in “sensitive and emotionally charged conversations” and are applied temporarily on a per-message basis.

This implementation is part of OpenAI’s efforts to improve ChatGPT’s response to situations involving mental and emotional distress, as previously announced. The company emphasizes that protecting young users and vulnerable individuals is a responsibility.

Despite this, many users see this approach as a restriction of their freedom of choice. Critics liken it to being “forced to watch television with parental controls constantly enabled in an environment where children are not even present.”

It appears that OpenAI’s new security-focused system will continue to be a topic of much debate in the tech world in the coming days.