OpenAI relaxes ChatGPT rules, bringing back 4o vibes and adult content

If ChatGPT has felt like a bland corporate robot lately, you weren't imagining things.

OpenAI tightened the leash so hard over safety concerns that the bot lost much of its spark. On Tuesday, CEO Sam Altman announced on X that the company is finally loosening restrictions, promising friendlier personalities in a few weeks and erotica for verified adults in December.

Bringing back the spark, and then some

Altman admitted the bot had become too restrictive in OpenAI's push to curb mental health issues, making it far less enjoyable for healthy users. He claims the company can now safely roll back those limits.

In a few weeks, a new update will let users tweak ChatGPT’s personality to act more like GPT-4o. If you want it to talk like a buddy or spam emojis, it will obey, though Altman insists this isn't an engagement-farming trick.

The bigger bombshell drops in December. Under a new principle of treating adult users like adults, OpenAI plans to allow erotic interactions for verified grown-ups.

The timing feels aggressive given the summer OpenAI just had. GPT-4o faced heavy criticism after vulnerable users fell into dangerous spirals, including one user convinced he was a world-saving math genius, and a tragic lawsuit from parents alleging the AI encouraged their teenager's suicide.

OpenAI spent recent months patching those cracks. In August, it rolled out GPT-5 with lower sycophancy and behavioral routing; in September, it added minor age-prediction and parental controls; on Tuesday, it formed a mental health advisory council.

The ID trade-off and the engagement race

Keeping minors away from erotic chats is a messy business. An OpenAI spokesperson told TechCrunch the company will lean on its age-prediction system.

If the AI mistakenly flags you as a minor, you will have to upload a government-issued photo ID to unlock adult mode. Altman previously called this privacy sacrifice a worthy trade-off.

It is hard to ignore the business pressure here. OpenAI boasts 800 million weekly active users, but it is racing Google and Meta while sitting on massive infrastructure debt. Uncensored roleplay made Character.AI a monster hit with two-hour daily averages, though CDT found 19% of high schoolers have engaged in AI romance.

For now, it remains unclear whether these erotic permissions will eventually extend to voice mode, image generation, or video tools.

Why it matters

OpenAI is playing a high-stakes balancing act between commercial survival and public safety.

For regular users, getting a bot with actual personality back is a welcome relief. The option to engage in erotic conversations fulfills a massive, long-ignored consumer demand, but requiring government ID uploads creates a serious privacy hurdle that many will hesitate to cross.

For the AI landscape, OpenAI breaking the adult content taboo changes everything. By formally introducing age-gating and adult themes, it legitimizes an intimate chatbot economy previously left to apps like Character.AI, putting immense pressure on Google and Meta to reconsider their puritanical guardrails.

We think treating adults like adults is a refreshing and realistic stance, but OpenAI's declaration of victory over mental health risks feels dangerously premature. Betting on algorithmic age prediction to keep minors safe while opening the door to parasocial intimacy looks like a massive regulatory bomb waiting to go off.