ChatGPT has taken the world by storm and helped spark an AI revolution, but new data suggests the honeymoon may be over.

According to The Washington Post, new research from Similarweb shows that ChatGPT users dropped by 9.7% in June, marking the first time the AI’s user base has dropped. The news throws doubt on the future of generative AI, at least in its current form.

Much of the disillusionment comes from the disconnect between what people expect from ChatGPT versus what it can actually deliver. Several major publications have tried using generative AI to rent articles, only to have issues with accuracy and credibility.

Generative AI models have also had issues with “hallucination,” a term the industry uses for those times when AI invents answers. Alphabet CEO Sundar Pichai recently admitted that all models hallucinate, and no one knows why.

As the Post points out, there are also issues with ChatGPT’s answers degrading over time, with some questioning if regulation is making AI dumber.

“If we continue to see an increase in output along the lines of ‘I am not able to answer that question because I am a chat bot,’ we will grow more concerned that regulation is removing power from ChatGPT,” Macquarie analyst Sarah Hindlian-Bowler wrote in a note to clients.