OpenAI’s planned adult mode for ChatGPT is starting to look less like a quirky product update and more like a serious test of how far AI companies can safely relax their guardrails.
The Verge reports that OpenAI’s delayed adult mode is expected to support adult-themed text conversations at launch, but not images, voice or video. The report says an OpenAI spokesperson described the feature’s intended output as “smut rather than pornography,” with the product focused on written adult-themed chats instead of visual explicit content.
What OpenAI Is Reportedly Planning
According to The Verge, the feature was first announced last October after CEO Sam Altman said OpenAI believed it had reduced enough safety risks to consider allowing “erotica for verified adults.” The feature had been expected sometime this quarter, but earlier this month OpenAI said it was delaying the rollout to focus on higher-priority work, and no new launch date has been publicly announced.
OpenAI has also publicly said it uses age prediction to identify whether an account may belong to someone under 18, and if a user is 18 or older, they can in some cases verify their age to turn off extra teen safety settings. That makes age checks part of the broader background here, even if the adult mode itself has not yet launched.
Why the Delay Matters
The Verge says the delay is tied not just to scheduling, but to internal concerns around safeguards, especially protecting minors and preventing unhealthy forms of emotional dependence on the chatbot. It reports that advisers warned OpenAI about the risk that adult mode could be accessed by children, and that moderation teams were also struggling with how to allow some adult-themed material while still blocking more harmful scenarios.
That tension is the real story. It is one thing for an AI company to say it wants to loosen restrictions for verified adults. It is another thing entirely to build a system that can reliably separate adults from minors, distinguish acceptable adult text from harmful material, and avoid pushing vulnerable users into unsafe territory. That is an inference based on the concerns described in The Verge’s reporting and OpenAI’s own explanation of how age prediction works.
The Bigger Debate Around AI and Boundaries
There is also a broader policy backdrop. OpenAI said last year that public input on its model behaviour showed significant support for allowing erotica for consenting adults, but the company added that it still had “more research and product work to do” before making changes. In other words, even where there may be user demand, OpenAI’s own public position suggested the company was not ready to flip the switch quickly.
The Verge also notes that keeping the feature text-only could make the legal and regulatory picture easier in some places than offering explicit images. But even if text is easier to moderate than images or video, that does not make the product simple. The challenge for OpenAI is not just whether adults want the feature. It is whether the company can build it without creating new safety and reputational problems at scale.
Why this matters for Australia
Questions around age verification, child safety and AI moderation are not just overseas policy issues. Australian regulators, parents, schools and platforms are already grappling with how online services should handle minors and sensitive content. A delayed product like this shows how difficult those problems remain even for one of the world’s biggest AI companies.
This is also another example of AI moving into areas that are socially and politically sensitive, not just technically impressive. The question is no longer only what chatbots can generate, it’s what kinds of boundaries companies can enforce when millions of users, including teenagers, are already on the platform.
For readers, the bigger takeaway is simple: AI companies may want more flexible adult features, but building them safely is proving much harder than announcing them.
Source: The Verge | OpenAI Help
