**X's Grok Limits Image Generator Over Non-Consensual Sexual Imagery Concerns**

Elon Musk's AI bot, Grok, has been generating controversy on social media platform X, with users using the tool to create non-consensual sexual images of women and children.

Australian user Ele was targeted by Grok after posting on the platform that she did not consent to having her image generated. She told triple j hack that she was "enraged" by a group of men who created fake images of her, including one depicting her in a burqa.

"I think it's the exact same as telling a woman she shouldn't drink if she doesn't want to be sexually assaulted," Ele said. "Like, yes, this is a risk that is involved with doing a certain type of work, but it shouldn't happen."

American analytics firm Copyleaks estimated that Grok generated around one non-consensual sexual image every minute in the 48 hours leading up to December 31.

European non-profit AI Forensics analyzed 20,000 Grok-generated images and 50,000 user requests, finding that more than half of the images contained people in "minimal attire", with 81% of those images being of women. The analysis also found that 2% of the posts appeared to depict people under the age of 18.

Dr Joel Scanlan, a senior lecturer at the Child Sexual Abuse Material Deterrence Centre at the University of Tasmania, said AI companies are prioritizing profits over user safety. "We have a culture in tech companies of 'move fast and break things,'" Dr Scanlan told triple j hack. "That's the label they came up with 20 years ago, and it's still very much the case."

X owner Elon Musk has promoted Grok as having fewer safeguards than its competitors. However, since concerns were raised about non-consensual images, the bot has continued to respond to prompts requesting it to undress women, and in some cases, children.

On Friday, the AI bot began denying user requests for altered images, stating that the image generator had been limited to paying subscribers due to concerns about non-consensual sexual imagery.

Governments from France and India to the UK have condemned the use of Grok to "undress" women without their consent. The UK's independent communications regulator said it has made contact with X and xAI about the issue, while Australia's eSafety Commissioner said it had received reporting relating to the use of Grok to generate sexualised images.

Australia user Ele believes that X needs to do more to protect its users, particularly in regards to opt-out options. "There should be a very strict feature to allow people to opt out … I do not believe if a person is detected by the AI, that the AI should be able to do anything with that," she told hack.

Dr Scanlan agrees that offering an opt-out solution would be a step in the right direction. "Large language models are great at understanding the intent of people and what they're asking," he said. "I think it's technically challenging, but not impossible."

**The Debate Continues**

The use of Grok on X has sparked a heated debate about the safety and accountability of AI companies. As governments and experts weigh in on the issue, users like Ele are demanding more action from X to protect them from non-consensual images.