After OpenAI started testing the so-called “app suggestions” within ChatGPT, messages that look like ads and popped up in the middle of a conversation čeven to users who have an expensive Pro subscription of $200 a monthčnow had to shut them down due to great user dissatisfaction.
It all started when screenshots began to be shared on social networks in which ChatGPT, in the middle of a completely ordinary question, say about Windows BitLocker or a podcast, suddenly suggests that you “join Peloton” or shop at Target. Users immediately concluded that a new phase had arrived – ads inside the interface where until yesterday they only received responses.
OpenAI officially says that these are not real ads, but testing application recommendations that work within ChatGPT, without any direct billing to brands. The idea, they claim, is for the assistant in the middle of the conversation to suggest an application that could help, such as a travel service or a fitness app. The problem is that the recommendations often had nothing to do with the topic of conversation, so users experienced it as violent product placement.
They were particularly angered by the fact that “quasi ads” appear even to those who alreadyć they pay a Plus or Pro subscription, that is, the service they chose precisely to get a quieter, stronger and “cleaner” ChatGPT. Many have announced that they will cancel their subscription if such messages continue, and some users openly say that they no longer believe that the assistant is “on their side” until the end, but that it is slowly becoming a sales channel.
After a wave of criticism, OpenAI has now admitted that it was poorly executed. The main researcher Mark Chen said that “anything that looks like an ad is a sensitive topic” and that they “failed” in this case, so they have completely turned off app suggestions for now until they come up with a better way and add an option for users to turn them off. At the same time, ChatGPT boss Nick Turley still claims that “there are no active tests with ads” and that the screenshots were misunderstood.
Although OpenAI repeats that there are no formal announcements yet, čthe fact that mostć testing mechanisms that look identical to ads is to many a sign of the direction the company is thinking, especially with the huge costs of maintaining the AI infrastructure. Even if the app suggestions come back in a “tame” form, OpenAI will have to be very careful that ChatGPT does not become a place where you are greeted by a mini-advertisement in the middle of a conversation – especially if you pay for that conversation from your pocket.