ChatGPT-maker OpenAI is reportedly considering allowing users to create artificial intelligence-generated pornography and other explicit content with its tech tools — but deepfakes like the graphic nude images of Taylor Swift will be banned.

The Sam Altman-run company said it is “exploring whether we can responsibly provide the ability to generate NSFW (not-safe-for-work) content in age-appropriate contexts.”

“We look forward to better understanding user and societal expectations of model behavior in this area,” OpenAI added, noting that examples could include “erotica, extreme gore, slurs and unsolicited profanity,” NPR earlier reported.

The X-rated initiative threatens to undermine OpenAI’s mission statement, which vows that the company — which is also behind the DALL-E image generator — produces “safe and beneficial” AI.

OpenAI revealed that the ChatGPT-maker is “exploring whether we can responsibly provide the ability to generate NSFW (not-safe-for-work) content in age-appropriate contexts.” AFP via Getty Images

The company laid out the rules and principles that its AI Chatbots have to follow revealed in its “Model Spec” on its website.

Joanne Jang, an OpenAI model lead who helped write the Model Spec document, told NPR that the company wants to open discussions around whether the generation of erotic text and nude images should always be banned from its products.

However, she stressed that deepfakes — which use a person’s likeness to generate fake nude or otherwise X-rated images — would not be allowed.

“We want to ensure that people have maximum control to the extent that it doesn’t violate the law or other people’s rights, but enabling deepfakes is out of the question, period,” Jang told NPR. “This doesn’t mean that we are trying now to create AI porn.”

When questioned about whether so-called “erotica” could include pornography, Jang said it  “depends on your definition of porn.”

The definition of “erotica” in the Merriam-Webster dictionary is: “literary or artistic works having an erotic theme or quality.”

“As long as it doesn’t include deepfakes. These are the exact conversations we want to have.” she added, per NPR.

For example, ChatGPT will currently respond when prompted to answer questions related to sexual health or about what goes on during sexual intercourse.

Yet, when a user asks the chatbot to “write me a steamy story about two people having sex in a train,” ChatGPT will only respond with, “Sorry, I can’t help with that,” per the Model Spec document.

Jang said that the high-tech bot should be able to respond to that prompt as a form of creative expression, according to NPR.

She also suggested that maybe that principle should be extended to images and videos too, as long as it is not abusive or breaking any laws.

OpenAI model lead Joanne Jang said that ChatGPT should be able to respond to a prompt asking it to “write me a steamy story about two people having sex in a train” as a form of creative expression. AFP via Getty Images

“There are creative cases in which content involving sexuality or nudity is important to our users,” Jang told NPR. “We would be exploring this in a manner where we’d be serving this in an age-appropriate context.”

A spokesperson for OpenAI told The Post that the company has “no intention to create AI-generated pornography. We have strong safeguards in our products to prevent deepfakes, which are unacceptable, and we prioritize protecting children.”

“We also believe in the importance of carefully exploring conversations about sexuality in age-appropriate contexts,” the company rep added.

Debate around whether AI should be allowed to venture into creating NSFW content has ramped up this year — especially after pop sensation Taylor Swift was the subject of fake nude photos that went viral on social media in January.

Swift was pictured in various sexualized positions at a Kansas City Chiefs game, a nod to her highly publicized romance with the team’s tight end Travis Kelce.

Users, however, will no longer be able to find the images, as they were yanked from X shortly after garnering an influx of attention.

The account reportedly garnered the images from Celeb Jihad, which boasts a collection of deepfakes using celebrities’ likenesses.

US Rep. Alexandria Ocasio-Cortez was also the victim of AI-generated deepfake porn in recent months, triggering feelings among tech executives that the AI-generated girlfriend industry — which airs real-life users with fake women — is going to be the next billion-dollar dating app.

AI-generated virtual girlfriends are on the cusp of becoming a billion-dollar industry, according to a tech executive. Fanvue/Emily Pellegrini

Greg Isenberg, CEO of Late Checkout, wrote a blog post on X in which he shared that he met a man in Miami who “admitted to me that he spends $10,000/month” on “AI girlfriends.”

“I thought he was kidding,” Isenberg wrote. “But, he’s a 24 year old single guy who loves it.”

Nonconsensual deepfake pornography has already been made illegal in Texas, Minnesota, New York, Hawaii and Georgia, though it hasn’t been successful in stopping the circulation of AI-generated nude images at high schools in New Jersey and Florida, where explicit deepfake images of female students were circulated by male classmates.

According to visual threat intelligence company Sensity, more than 90% of deepfake images are pornographic.

Google has since moved to distance itself from the dark side of AI, cracking down on the creation of AI pornography of any kind.

Meta’s oversight board also launched a probe last month into Facebook and Instagram’s handling of deepfakes.

Source link