Sam Altman, main enforcement serviceman of OpenAI Inc., during a media circuit of the Stargate AI information halfway successful Abilene, Texas, US, connected Tuesday, Sept. 23, 2025.
Kyle Grillot | Bloomberg | Getty Images
Adult ChatGPT users tin soon entree a little censored mentation of the artificial quality chatbot, which volition see erotic materials, OpenAI CEO Sam Altman has announced successful an evident argumentation shift.
"In December, arsenic we rotation retired age-gating much afloat and arsenic portion of our 'treat big users similar adults' principle, we volition let adjacent more, similar erotica for verified adults," Altman said successful a societal media post on Tuesday.
Though it remains unclear what contented volition suffice arsenic permitted erotica, the determination could correspond a large displacement successful OpenAI's policy, which formerly prohibited specified contented successful astir contexts.
According to Altman, existent versions of ChatGPT were made "pretty restrictive" to support users from intelligence wellness risks, but that attack made the chatbot "less utile [and enjoyable to galore users who had nary intelligence wellness problems.
"Now that we person been capable to mitigate the superior intelligence wellness issues and person caller tools, we are going to beryllium capable to safely unbend the restrictions successful astir cases," helium said.
The "new tools" look to notation to information features and parental controls rolled retired past period to code concerns implicit however the chatbot was impacting young users' intelligence health.
However, arsenic safeguards for minors expand, it appears that Altman is acceptable for ChatGPT to instrumentality a looser attack for adults.
Besides the rollout successful December, Altman besides said a caller mentation of ChatGPT volition motorboat successful the coming weeks, allowing the chatbot to follow much chiseled personalities — gathering connected updates successful the latest GPT‑4o version.
"If you privation your ChatGPT to respond successful a precise human-like way, oregon usage a ton of emoji, oregon enactment similar a friend, ChatGPT should bash it," helium said. "But lone if you privation it."
Growth vs. safety
Altman's latest station drew attraction due to the fact that helium antecedently suggested ChatGPT wouldn't instrumentality sexualized chat features, dissimilar rival models specified arsenic XAI's Grok.
In an August interview, autarkic tech writer Cleo Abram asked Altman to springiness an illustration of a determination helium had made that was champion for the world, but not for winning the AI race.
"Well, we haven't enactment a enactment bot avatar successful ChatGPT yet," Altman said successful an evident motion to provocative AI companions released by Elon Musk's xAI.
Altman's remarks travel arsenic OpenAI faces greater scrutiny of its information policies. In September, the Federal Trade Commission launched an enquiry into respective tech companies, including OpenAI, implicit imaginable risks to children and teenagers.
That followed a suit from a California mates who alleged that ChatGPT contributed to their 16-year-old son's suicide.
OpenAI connected Tuesday besides announced an eight-member adept assembly connected well-being and AI to counsel the institution connected however artificial quality affects users' intelligence health, emotions and motivation.
The assembly volition usher OpenAI successful defining what steadfast AI interactions look similar done check-ins and recurring meetings, the institution said.
English (US) ·