Elon Musk's Grok sparked a backlash after its AI image generator was used to generate non-consensual sexual images of real people, including minors.
Over the past week, some X users have used Grok to digitally undress people in photos, generating fake images in which the AI model shows the subjects more skin, wears bikinis, or changes their body position.
Some of the requests are consensual, such as an OnlyFans model asking Grok to remove his clothes. But others urged Grok to “remove the clothes” from images of adults who are not themselves. Some of these images include minors, according to screenshots posted by concerned users on social media platforms and multiple examples seen by Business Insider.
XAI's “Terms of Use” policy prohibits “portraying the likeness of any person in a pornographic manner” and “sexualization or exploitation of children.”
When asked for comment, xAI sent an automated email response that did not address the issue.
French authorities are investigating a rise in AI-generated deepfakes from Grok, the Paris public prosecutor's office told Politico. Distributing non-consensual deepfakes online is punishable by two years in prison in France.
In a letter to the chief compliance officer of Company X's India operations, India's Ministry of Electronics and Information Technology described reports of users distributing “images and videos of women in a derogatory or vulgar manner with the purpose of degrading and degrading women.”
The ministry asked X to undergo a “comprehensive technical, procedural and governance-level review” and remove content that violates Indian law.
Alex Davies-Jones, the UK's minister for violence against women and girls, has implored Elon Musk, CEO of X's parent company xAI, to take action regarding AI images.
“If you care so much about women, why do you allow X users to exploit them?” she wrote. “Glok can undress hundreds of women every minute, often without the knowledge or consent of the people in the images.”
Davis-Jones also referred to the UK's proposal to make it a crime to create and distribute sexually explicit deepfakes.
In response to an X user reporting a screenshot showing Grok creating sexual images of minors, Grok's official account said the company had “identified a flaw in our security measures and is urgently fixing it.” However, it is unclear whether Grok's response was considered by xAI or simply generated by AI.
Grok's official account responded in a separate thread: “There have been isolated cases where users have requested and received AI images depicting minimally clothed minors, such as the example you mentioned.” “xAI has safeguards in place, but improvements are underway to completely block such requests.”
While Musk has touted Grok's NSFW capabilities, deepfakes are an ongoing concern and moderation challenge for AI companies.
In August, Grok's image and video generator Imagine launched a “Spicy” mode that allows users to create AI-generated pornographic images of women. The “spicy” option wasn't available for photo uploads, but users could enter custom prompts like “take off your shirt.”
The “undress” Grok trend skyrocketed after Wired reported on Dec. 23 that OpenAI's ChatGPT and Google's Gemini AI models were being used to generate images of real women in bikinis from fully clothed photos.
People vary in their ability to counter self-created AI deepfakes.
In the United States, take-it-down laws prevent non-consensual deepfakes, but the scope varies by age and body part displayed. For adults, the law only covers deepfakes that show genitals or sexual activity. The law is tougher on minors and covers deepfakes that are intended to “abuse, humiliate, harass, or degrade” or “stimulate or gratify the sexual desires of any person.”
Some states have also passed stricter laws regarding the spread of deepfakes.
The creation of deepfakes by AI raises more complex liability issues, but Section 230 of the Communications Decency Act of 1996 primarily exempts online platforms from liability for content posted by their users.
Speaking to Business Insider in August, Alison Mahoney, a technology-facilitated abuse lawyer, questioned whether AI-generated tools could “strike away immunity” from viewing platforms as creators.
“We need clear legal avenues to hold platforms accountable for wrongdoing,” Mahoney said.
