In October, ChatGPT maker OpenAI announced it would be opening the floodgates for “mature apps.”
“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 at the time.
“As part of our ‘treat adult users like adults’ principle, we will allow even more, like erotica for verified adults,” Altman added.
Five months later, an “adult mode” chatbot that’s willing to breach topics that have been off-limits on ChatGPT so far — a move characterized by critics as a way to boost revenue in light of some disastrous financials — remains nowhere to be seen.
And as the Wall Street Journal reports, the subject is still sending a shiver down the spines of company advisors, who are wary of the many potential dangers of letting OpenAI’s already-hooked customers engage in intimately-charged conversations.
In fact, many staffers and executives were reportedly blindsided by Altman’s promise in the first place, making an imminent launch out of the question.
Despite plenty of concerns and internal debates over the risks, from users growing too emotionally attached to compulsive use, OpenAI is reportedly still forging ahead. (The company did admit earlier this month that adult mode’s launch would be delayed as other products were being prioritized.)
Plenty of glaring security issues remain, with inside sources telling the WSJ that its new age-prediction system has been misclassifying minors as adults 12 percent of the time. While that may not sound like much, multiplied by ChatGPT’s enormous user base, millions of underage children could be accessing inappropriate chats.
In an effort to keep nonconsensual sexual images off the platform — something competitor Elon Musk’s xAI has been unsuccessfully grappling with — OpenAI is playing it relatively safe by restricting spicy conversations to just text.
It’s also trying to control the narrative by painting its new feature as a way to generate something you’d find in romance novels. A spokeswoman told the WSJ that its erotica chats were more akin to “smut rather than pornography.”
The spokeswoman also assured that users will be encouraged to seek relationships in the real world.
But given the shaky track record of implementing effective guardrails and moderating explicit content — while Altman claims “serious mental health issues” are no longer a problem for OpenAI, a wealth of data suggests otherwise — it remains to be seen how OpenAI’s “adult mode” will fare.
If xAI’s Grok is anything to go by, the risks are considerable. Users have been using the chatbot to unclothe images of real people, resulting in a wave of nonconsensual pornographic images flooding the largely unmoderated social media site.
Its ongoing struggles with child sex abuse material (CSAM) culminated in a lawsuit on behalf of three teens, including two minors, which was filed in the Northern District of California today. The plaintiffs accuse xAI of fostering an environment that allowed for the spread of CSAM.
There have also been countless instances of users forming intense relationships with AI chatbots. Underage users are particularly vulnerable, often developing strong bonds without their parents’ knowledge.
In extreme cases, the phenomenon has been linked to a string of tragic teen suicides, culminating in several high-profile lawsuits aimed at OpenAI and its competitors.
In short, OpenAI is painfully aware of the risks in rolling out its “adult mode” feature.
According to the WSJ, though, it’s looking to launch in a month or so.
“We still believe in the principle of treating adults like adults,” the company told the newspaper, “but getting the experience right will take more time.”
More on OpenAI and smut: OpenAI Says It Will Move to Allow Smut
