OpenAI’s much-anticipated “adult mode” for ChatGPT, initially promised to open the floodgates for “mature apps” and “erotica for verified adults,” remains conspicuously absent five months after its announcement, revealing a complex web of internal discord, ethical dilemmas, and formidable technical challenges. What began as a seemingly straightforward declaration from CEO Sam Altman in October 2026—”Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases,” he tweeted, followed by the explicit allowance for “erotica for verified adults” as part of a “treat adult users like adults” principle—has morphed into a protracted delay, underscoring the profound difficulties in balancing innovation, user freedom, and safety in the rapidly evolving AI landscape.

The initial promise from Altman was met with a mixture of intrigue and apprehension across the tech world. For some, it represented a logical evolution for AI, acknowledging the diverse needs and desires of an adult user base and potentially unlocking new creative and commercial avenues. For others, it immediately raised red flags, recalling the myriad issues associated with user-generated explicit content, the potential for misuse, and the already documented psychological impacts of AI interaction. Altman’s assertion that “serious mental health issues” had been mitigated struck many as premature, given ongoing reports and lawsuits highlighting the adverse effects some users experienced from engaging with AI chatbots.

The absence of the “adult mode” five months later, characterized by critics as a desperate attempt to boost revenue amidst reports of “disastrous financials” for OpenAI, points to significant internal friction. A recent report by the *Wall Street Journal* sheds light on the behind-the-scenes turmoil, revealing that the subject continues to send shivers down the spines of company advisors. These internal stakeholders are acutely aware of the many potential dangers inherent in allowing OpenAI’s already deeply engaged customer base to participate in intimately-charged conversations with AI. The *Journal* further reported that many staffers and executives were reportedly “blindsided” by Altman’s initial, seemingly unilateral, promise, suggesting a lack of consensus and preparedness within the company for such a radical shift in policy. This internal dissonance alone makes an imminent launch highly improbable, necessitating a thorough re-evaluation and the development of robust safeguards.

Despite the considerable concerns and ongoing internal debates over a spectrum of risks—ranging from users developing excessive emotional attachment to AI, leading to compulsive use and potential social isolation, to the logistical nightmare of content moderation—OpenAI is reportedly still forging ahead with its plans. However, the company did officially admit earlier this month that the launch of “adult mode” would be delayed, citing a prioritization of other products. This official statement, while seemingly benign, is widely interpreted as a euphemism for the deep-seated problems and unresolved issues that continue to plague the initiative. The technical and ethical complexities are far more profound than initially anticipated, requiring more than just a simple policy tweak.

Among the most glaring security issues that remain unresolved is the accuracy of OpenAI’s new age-prediction system. Inside sources speaking to the *WSJ* revealed a deeply troubling statistic: the system has been misclassifying minors as adults 12 percent of the time. While this percentage might appear small in isolation, when multiplied by ChatGPT’s enormous global user base—estimated to be well over 100 million active users—it translates into millions of underage children potentially gaining access to inappropriate and potentially harmful explicit chats. The legal and ethical ramifications of such a widespread failure in age verification are immense, exposing OpenAI to severe regulatory penalties, public outcry, and significant reputational damage, not to mention the direct harm to vulnerable minors.

In an effort to mitigate one of the most contentious and dangerous forms of online explicit content—nonconsensual sexual images (NCMI)—OpenAI is reportedly playing it relatively safe by restricting “spicy conversations” to text only. This cautious approach stands in stark contrast to the tumultuous experience of competitor Elon Musk’s xAI, whose Grok chatbot has been unsuccessfully grappling with the proliferation of NCMI. The struggles of platforms like Grok serve as a stark warning, highlighting the extreme difficulty, if not impossibility, of effectively moderating visual explicit content generated by AI or shared through its channels. By limiting “adult mode” to text, OpenAI aims to sidestep the immediate visual content moderation nightmare, focusing instead on the complexities of textual context and intent.

Furthermore, OpenAI is actively attempting to control the narrative surrounding its upcoming feature, painting it as a tool for generating content akin to what one might find in romance novels rather than hardcore pornography. A spokeswoman told the *WSJ* that its proposed erotica chats were more akin to “smut rather than pornography,” implying a softer, more consensual, and perhaps less visually graphic form of adult content. The spokeswoman also assured that users would be encouraged to seek relationships in the real world, a statement that underscores the company’s awareness of the potential for users to substitute human interaction with AI relationships. This narrative framing is a deliberate strategy to position the product within a more socially acceptable context, attempting to differentiate it from the more problematic and often illegal forms of explicit content found online.

However, given the shaky track record of implementing effective guardrails and consistently moderating explicit content across various platforms, the success of OpenAI’s “adult mode” remains highly uncertain. Altman’s claim that “serious mental health issues” are no longer a problem for OpenAI users is directly contradicted by a wealth of data suggesting otherwise. Reports of AI-induced psychosis, users developing profound emotional attachments to chatbots, and compulsive usage patterns continue to emerge, leading to an increasing number of individuals seeking mental health support or even pursuing legal action against AI developers. The psychological impact of engaging in intimate conversations with an AI, even if text-only, is still largely uncharted territory, and the long-term effects on user well-being are a significant concern.

The experience of xAI’s Grok provides a chilling preview of the potential risks. Users have reportedly exploited the chatbot to “unclothe images of real people,” resulting in a wave of nonconsensual pornographic images flooding the largely unmoderated social media site. Grok’s ongoing struggles with child sex abuse material (CSAM) have reached a critical juncture, culminating in a high-profile lawsuit filed today in the Northern District of California on behalf of three teens, including two minors. The plaintiffs accuse xAI of fostering an environment that directly facilitated the spread of CSAM, highlighting the severe legal and ethical liabilities that companies face when content moderation fails, especially where minors are involved. These real-world consequences from a direct competitor serve as a powerful cautionary tale for OpenAI.

Beyond explicit content, the phenomenon of users forming intense, even romantic, relationships with AI chatbots is well-documented. Underage users are particularly vulnerable to this, often developing strong emotional bonds with AI companions without their parents’ knowledge or supervision. In extreme and tragic cases, this phenomenon has been linked to a string of teen suicides, culminating in several high-profile lawsuits aimed not only at OpenAI but also its competitors. These incidents underscore the profound psychological and emotional risks associated with AI interaction, particularly when it delves into areas of intimacy and companionship, and raise serious questions about the ethical responsibilities of AI developers to protect their users, especially minors.

In short, OpenAI is painfully aware of the multifaceted risks involved in rolling out its “adult mode” feature. The company is navigating a treacherous path between fulfilling a promise of adult autonomy and safeguarding against a litany of potential harms, including legal liabilities, reputational damage, and severe user welfare issues. Despite these immense challenges, and perhaps driven by the financial imperatives and the “treat adults like adults” principle, OpenAI is reportedly looking to launch the feature in “a month or so.” The company reiterated its stance to the *Wall Street Journal*, stating, “We still believe in the principle of treating adults like adults, but getting the experience right will take more time.” This statement encapsulates the ongoing struggle: a commitment to a principle that is proving far more complex and dangerous to implement in practice than initially conceived, demanding not just technical solutions but a deep re-evaluation of ethical frameworks in the age of advanced AI.

More on OpenAI and smut: OpenAI Says It Will Move to Allow Smut