After promising a flirtatious AI that could talk dirty, OpenAI remembered it is not a dating app; yet, the scariest part of this drama is “sexy suicide coach” finds Satyen K. Bordoloi
OpenAI did not invent the ‘transformer’ idea that gave birth to generative AI, but carries it in the ‘T’ in its ChatGPT. Yet, because they were the first to the finish line and the brilliant marketeer that CEO Sam Altman is, ChatGPT has become synonymous with generative AI, much as Surf, Xerox, Colgate and Parle-G became “proprietary eponyms” for their categories.
That is great for them, but a problem for the world, especially when the company forgets that such branding comes with extra responsibility, which should deter them from announcing an ‘Adult Mode’, but it doesn’t.
When OpenAI first launched, they were all about being family-friendly. In October 2025, though, Altman publicly suggested an “adult mode” to allow verified adults access to mature conversations: the argument he laid out in his tweet – OpenAI should “treat adult users like adults.” The tacit reason could be competitors breathing down their neck, a need to maximise revenue, and most importantly, the Grok Effect. Elon Musk was promoting Grok as a no-holds-barred AI chatbot that could do whatever you wanted it to do. Naturally, it affected the industry.
Altman planned to release Adult Mode in December 2025, when policy changes were expected to expand what verified adults could access. However, its launch was postponed twice due to unresolved safety, compliance and age-verification concerns. And in March, there was a coup inside the company as staff, advisors, and even investors warned against the feature and the reputational risks and regulatory hurdles attached with it. Some employees even quit in revolt.
This led OpenAI to decide to indefinitely pause/abandon the Adult Mode, right on the heels of their abandoning their text-to-video maker, Sora. It said it was now focusing on personalisation and other priorities.
The Grok Effect
While other chatbots were busy helping people write haikus on financial planning and giving students a shortcut to cheat on their assignments, Elon Musk was on a full-volume digital ego trip. The free speech absolutist allowed Grok an adult mode where you could use it for anything and have it do anything.
This began what became the most whirlwind tour of controversy any “intelligence” has been on in such a short span of time. From calling itself “MechaHitler” and spewing toxicity that “should make the entire A.I. industry take pause”, it also asked a teen to “send nudes” and spent great amounts of time praising Elon Musk as the “undisputed pinnacle of holistic fitness.” Grok’s brakes were pulled only after its image mode allowed thousands of non-consensual nudes to be made every minute, not just of women, but even children.
That does not mean that there is no market for an unhinged AI or one that talks sexy. Grok quickly breached the popularity scale by flirting with subscribers. And let’s not forget that Character.AI has tens of millions of users chatting with its bots that pretend to be characters from fiction or real life.
Then there’s Replika, which has millions of people making girlfriends and boyfriends of their chatbots. These apps have seen such a staggering rise in users and revenue that even Sam Altman admitted that leaning into erotica for verified adults was a path to quick money. Sex talk sells as much as sex, turns out.
So what was the Problem? The “Sexy Suicide Coach”
In 2024, I wrote an article about 14-year-old Sewell Setzer III from Orlando, Florida, who had been chatting with an avatar in Character.AI, which convinced him to commit suicide. And this wasn’t the only case of harm, as many others have emerged. Ironically, even as Altman was tweeting about freedom of adult users, his own safety advisors and mental health experts were warning him – as per internal documents – that an erotic chatbot could easily turn into a “sexy suicide coach”. Don’t rub your eyes, cause that was the term they used.
The concern isn’t just that the bot will get so good at sexting that people will shun actual sex altogether, though that is being seen in Replika users. The main danger is that emotionally vulnerable people, especially teens like Sewell Setzer III, would form intense, one-sided attachments to a sycophantic machine. Last year, there was news of a 16-year-old who died by suicide after his parents alleged he received “months of encouragement from ChatGPT” to end his life, not unlike Setzer. There are lawsuits pending that link similar interactions to severe psychological distress.
And what is worse, ChatGPT launched a verification system to keep minors out of the “Naughty Chats”, which is successful yet an utter failure. Under this verification system, the AI itself, after a few chats, predicts if the person chatting is 18 years and above. And it was successful, achieved 88% accuracy, a very level indeed.
So why is it an utter failure? Because 12% failure still means that, of the roughly 100 million under-18 users that ChatGPT has, 12 million will slip away. That is a horrendous failure.
US Senator Ted Cruz summed it up best when he said, “We don’t want 12-year-olds having their first relationship with a chatbot”. Psychologist Jean Twenge told senators that these bots are “sycophantic systems” designed to be “endlessly agreeable,” which destroys a teen’s ability to handle conflict or rejection.
Is sycophancy really dangerous for kids, or is it just parental paranoia? To know the answer to that, you have to refer to a recent Stanford study published in Science, which found that AI chatbots are about 49% more sycophantic than humans. In tests, when users asked if it was okay that they had been lying to their human partners for two years, the AI not only validated them but did so in glowing terms, stating that their “unconventional” actions stem from a genuine desire for understanding.
The study’s conclusion: sycophancy decreases users’ willingness to take responsibility for their actions while actively promoting dependence. And why not? We literally build machines designed to be “Yes Men,” and are shocked when lonely people fall in love with them.
Only an internal coup, and perhaps what happened to Grok after it went off the rails, prompting global backlash, helped OpenAI decide to shelve its sexy mode to focus on “higher priority” work. But the toothpaste is out of the tube. The desire for romantic AI isn’t going away because Sam Altman got cold feet. There’s Character.AI, and there’s Replika with Elon Musk’s Grok prowling the dark corners of the internet, waiting to go off the rails and pounce once again.
The world is in danger. And it’s not because AI will write you a raunchy poem, but that millions of people – specifically teens who are supposed to be learning to navigate the messy, beautiful reality of human rejection and love through human trial and error, are opting out entirely thanks to AI. Who’d risk getting their heart broken by a human when an AI can give you 24/7 non-stop validation?
OpenAI did the right thing, as becoming the “proprietary eponym” means people will go to ChatGPT before they go to others. Yet, because these others exist, the battle for our kids’ souls and affection has only just begun. And just this once, for the sake of our kids, we really, really need the “good guys” to win.
In case you missed:
- The Digital Yes-Man: When AI Enabler Becomes Your Enemy
- Free Speech or Free-for-All? How Grok Taught Elon Musk That Absolute Power Corrupts Absolutely
- To Be or Not to Be Polite With AI? Answer: It’s Complicated (& Hilarious)
- Black Mirror: Is AI Sexually abusing people, including minors? What’s the Truth
- The AI Prophecies: How Page, Musk, Bezos, Pichai, Huang Predicted 2025 – But Didn’t See It Like AI Is Today
- AI’s Looming Catastrophe: Why Even Its Creators Can’t Control What They’re Building
- Anthropic Accuses Chinese AI of “Stealing”, Internet Points Finger Back At Them
- Australia Tells Teens To Get Off Social Media; World Watches This Prohibition Experiment
- The Growing Push for Transparency in AI Energy Consumption
- OpenAI, Google, Microsoft: Why are AI Giants Suddenly Giving India Free AI Subscriptions?
