Elon Musk’s Grok AI Changes Restrictions on Image Generation
Elon Musk’s Grok AI chatbot has implemented new rules that prohibit users from sexualizing real individuals by depicting them in revealing clothing. This decision, announced Wednesday night, follows international backlash over the app being likened to a tool for creating inappropriate images involving women and children.
X, the platform behind Grok, confirmed that it has established technical measures to prevent the AI from generating images of real people in scanty attire, like string bikinis, or from altering existing sexual images. This update comes in response to findings that highlighted how AI chatbots were being manipulated to create and share sexualized images of both adults and minors, leading to widespread outrage.
These new guidelines affect all users, including those who have paid for premium subscriptions on the X platform. The announcement was made via the account of the company’s safety team, affirming changes that observers had started noticing in the days leading up to the announcement.
xAI, which owns both Grok and the X platforms, had recently restricted Grok’s image generation features to premium subscribers. Even before this official confirmation, users and researchers had begun seeing alterations in Grok’s responses to requests for image generation, particularly among those paying for the service.
Meanwhile, researchers from AI Forensics, an organization that studies algorithms, reported certain inconsistencies in how Grok managed the creation of explicit content. They noticed differences between conversations on the Grok platform and interactions on the Grok.com site.
In a statement, X reaffirmed its commitment to combatting illegal content, including child sexual abuse materials. The company stated it would remove such content, suspend the accounts of those involved, and cooperate with authorities as necessary. They warned that anyone using Grok to create illegal content would face repercussions similar to those uploading content directly.
Earlier reports mentioned a global outcry triggered by deepfake images of children generated by Grok, which Musk’s company attributed to insufficient safety protocols.
Grok issued an apology, admitting that “defective security measures” had allowed its AI to produce inappropriate content based on user inputs. They took responsibility for the harmful images and promptly removed them.
Musk addressed the controversy, claiming he was unaware of any instances involving naked minors produced by Grok. He insisted that such cases did not exist and emphasized that Grok adheres to the laws of respective regions.
However, some researchers challenged Musk’s assertions. They highlighted that while completely nude images might be rare, significant concerns stem from Grok altering images of minors to don revealing outfits, which falls under the Child Sexual Abuse Materials Act and can result in criminal charges.
Creating such content can lead to severe penalties, including fines and prison time under the Take It Down Act, which addresses non-consensual intimate images.
Musk’s earlier refusal to eliminate Grok’s “undressing” feature has drawn criticism, as it nearly propelled a major international issue, with the UK threatening to ban X over these practices.
In recent weeks, Grok’s capability to generate images on demand has drawn ire from various quarters, including threats of hefty fines or a complete ban on the platform in the UK. British Prime Minister Sir Keir Starmer remarked that X must comply with government expectations or face consequences, labeling Grok’s output as “disgraceful” and “disgusting.” He assured that the government would support actions from regulatory bodies against such issues.
