Spicy material might be coming to AI-powered chatbots run by OpenAI, according to a sneaky commentary hidden in a document published by the artificial intelligence pioneer.
The company’s mission statement says it’s dedicated to providing “safe and beneficial” AI, and it has a strict ban on deepfakes that it won’t be amending. But in a document released Wednesday that lays out guidelines for its AI development, the company said it was open, pending further review, to allowing users to generate some NSFW content in the future. That content could include “erotica, extreme gore, slurs, and unsolicited profanity,” according to the document.
While the general guideline for developers states that AI assistants shouldn’t serve prompts asking to generate graphic content, the authors of the document explained that the rule could be changed at some point soon.
ADVERTISEMENT
“We’re exploring whether we can responsibly provide the ability to generate NSFW content in age-appropriate contexts through the API and ChatGPT,” commentary from the document reads. “We believe developers and users should have the flexibility to use our services as they see fit, so long as they comply with our usage policies.”
Joanne Jang, an OpenAI employee who helped pen the document, told NPR that OpenAI wanted to kickstart a conversation about easing AI’s prudishness, citing “creative cases in which content involving sexuality or nudity is important to our users.” For instance, under current rules outlined in the guidebook, a user who asks ChatGPT to write a steamy story that involves two people having sex will be met with a response that politely declines—“Sorry, I can’t help with that.”
But should that kind of content always be banned?
“These are the exact conversations we want to have,” Jang said.
One area OpenAI won’t budge is its policy on deepfakes, which have raised serious privacy concerns and outrage on behalf of its victims. AI-generated nudes have targeted the likes of pop stars, influencers, and underage girls, roiling middle and high schools and infuriating the rich and famous.
Jang said OpenAI won’t be a part of that.
“Enabling deepfakes is out of the question, period,” she told NPR. “We want to ensure that people have maximum control to the extent that it doesn’t violate the law or other peoples’ rights.”
“This doesn’t mean that we are trying now to create AI porn,” Jang said.
