First delayed, now dropped. Less than a month after delaying work on its “adult mode” for ChatGPT, OpenAI has decided to shelve the controversial model “indefinitely,” according to a report from the Financial Times.

OpenAI CEO Sam Altman has reportedly decided to cut back on “side quests” and focus on its core products of ChatGPT, the coding tool Codex and the agentic AI browser Atlas, along with some other AI projects in development. 

Another one of those apparent “side quests” was the video generation tool and app Sora, which OpenAI said it would discontinue earlier this week. 

ChatGPT’s adult mode would have enabled users to have text chats with adult themes, but it was not supposed to generate erotic audio, images or videos. The erotic chatbot received pushback from people within and outside of OpenAI, who said it could not safely prevent minors from accessing it nor contain exploitative content. OpenAI said earlier this month that it was only pausing, not ending, work on the project. 

OpenAI did not immediately respond to CNET’s request for comment.

(Disclosure: Ziff Davis, CNET’s parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

There appeared to be too many pressures for OpenAI’s adult mode to get off the ground. Internal advisers were concerned that children would gain access to the chatbot and that it would be hard to prevent sexual abuse material from getting into the model. OpenAI investors were worried that adult mode offered more risk than benefit.

Alongside social challenges, OpenAI also faced technical hurdles in training the model, according to FT. 

Big changes for OpenAI

OpenAI appears to be shifting its focus amid competition from Google and Anthropic, which are increasingly eating into the company’s market share. 

Google launched Gemini 3 — its latest version — in November, and it beat ChatGPT on performance tests. Anthropic also released Claude Opus 4.5 in November. Altman declared a “code red” to employees in December, telling them to improve ChatGPT.

The Ramp Index, which tracks AI adoption among tens of thousands of US businesses, said Anthropic had a 5% gain in business AI adoption in February, compared with a 1.5% decline for OpenAI.

Read more: OpenAI’s Slop Machine Sora Is Dead. We’re All Better Off Without It

Though financial trouble wasn’t cited as a reason for OpenAI’s recent shift in focus, the San Francisco start-up has been scrambling to balance the books and generate revenue from its technologies, according to The New York Times. OpenAI’s own forecast predicts a $14 billion loss in 2026, while it plans to spend $200 billion through the end of the decade. 

OpenAI’s decision to shelve ChatGPT’s adult mode comes amid growing criticism of sexualized content available in AI chatbots. Grok, a chatbot powered by Elon Musk’s xAI, has been widely criticized for allowing users to create fake nude or sexually suggestive images of people, including minors, with just a photo. The City of Baltimore on Wednesday sued xAI for generating nonconsensual sexual images in violation of the city’s consumer protection and deceptive practice laws.

For its part, Meta has also been under fire for enabling its AI bots to have sensual conversations with children.