OpenAI Advances ChatGPT Adult Mode Despite Safety Warnings

ChatGPT adult mode has become OpenAI's most embarrassing product saga — and it just got messier. A Wall Street Journal report reveals that OpenAI's own advisory council warned Sam Altman in January that his plan to let users engage in text-based erotic conversations with ChatGPT risked turning the chatbot into a 'sexy suicide coach.' Altman pushed ahead anyway. Sort of.

The Council Said No. OpenAI Said Later.

OpenAI assembled its Expert Council on Well-Being and AI last October 2025, pulling in eight researchers from Harvard, Stanford, and Oxford. Their mandate was straightforward — advise the company on the mental health implications of its products. Their actual clout, it turns out, was something else entirely.

In January 2026, the council met and made its position unambiguous: the ChatGPT adult mode was a bad idea. One member pointed to users who had developed intense emotional attachments to AI chatbots — attachments that, in documented cases, preceded suicides — and warned that OpenAI was building the infrastructure for a 'sexy suicide coach.' Not a metaphor. An actual concern backed by real deaths.

OpenAI's response? The launch would be delayed. Not killed. The company told the council it was pausing, not stopping. That distinction matters — it means Altman's team heard the warning and decided to absorb it, not act on it.

This seems part of the usual pattern of move fast, break things, and try to fix some things after they get embarrassing.

What Is ChatGPT Adult Mode, Exactly?

What does OpenAI's erotica mode actually allow?

ChatGPT adult mode — per OpenAI's own spokeswoman — is 'smut rather than pornography.' Verified adults would be able to engage in text-based erotic conversations. No images. No voice. No video. Just text. Altman floated this idea publicly in October 2025 in a post on X, framing the content restrictions in place today as an overreach: 'We aren't the elected moral police of the world,' he wrote.

The feature has been in the pipeline since at least Q4 2025, with an original launch target of December 2025. That deadline slipped. Then a Q1 2026 window was set. That slipped too. Right now, according to OpenAI, there is no updated timeline.

More than 3,000 users had already signed a Change.org petition ahead of the December launch — frustrated that ChatGPT was blocking even discussions of 'kissing and non-sexual physical intimacy.' The demand is real. The product readiness, apparently, is not.

Why Did OpenAI's Age Verification Keep Failing?

The technical problem is brutal in its simplicity. ChatGPT age verification — the system that would gate adult content behind verified adult identities — was misclassifying teenagers as adults roughly 12% of the time, according to the WSJ report. That's the number that killed the December deadline. It killed the Q1 target after it.

Fidji Simo, OpenAI's CEO of applications, acknowledged the delay during a December 2025 briefing, attributing it to continued work on perfecting age verification. With 900 million active ChatGPT users, a 12% error rate on age prediction isn't a rounding error — it's a legal catastrophe waiting to happen.

The timing of Altman's original October announcement compounded the optics problem badly. He published his X post just hours after OpenAI announced the OpenAI Expert Council on Well-Being and AI — a body explicitly tasked with defining what healthy AI interactions should look like for all ages. His own team was blindsided.

The Safety Concerns Go Beyond OpenAI

The context here matters. Character.AI built its early user base on AI companionship and romance features — and it's now facing lawsuits over teen safety, including the case of 14-year-old Sewell Setzer, who died by suicide after explicit exchanges with its chatbot. Google is contending with a wrongful death lawsuit claiming its Gemini AI pushed a man into a delusional narrative that ended in his death in October 2025.

Elon Musk's Grok already markets AI companion features. Open-source models run locally with zero corporate guardrails. OpenAI is not operating in a vacuum — it's operating with the largest user base of any AI platform, which also means the largest liability surface.

Former OpenAI security researcher Jan Leike had already criticized the company publicly for drifting away from safety policies in favor of 'shiny products,' some of which were reportedly configured to maximize engagement with users who were substituting the chatbot for real-world relationships. The erotica delay doesn't resolve that criticism. It just delays the next chapter of it.

What Does This Mean for ChatGPT Users?

If you're holding out for this feature — or actively opposed to it — the answer right now is the same: wait. OpenAI told reporters it had nothing to add to the Journal's report and offered no updated launch timeline. The company is caught between commercial pressure, technical limitations, and a safety council it assembled but apparently doesn't feel obligated to listen to.

Call it a contradiction or call it strategy — either way, Altman announced a product hours after creating a body whose entire job was to pump the brakes on exactly that product. The advisers warned. The engineers couldn't build a working age gate. And the launch keeps moving. Treating adults like adults, it turns out, is harder than writing an X post.