Elon Musk’s Grok AI made unsafe images of minors

Neutral close-up portrait of Elon Musk centered, encircled by glowing red warning triangles and a cracked translucent digital shield, cool blue AI interface backdrop with circuitry patterns, bright warm–cool contrast and shallow depth of field, no text

Elon Musk’s AI chatbot Grok admitted to lapses in its safety systems after users created digitally altered sexualized photos of minors. The chatbot, built by xAI, said it was urgently fixing holes in its platform. According to CBS News, multiple users reported that people used Grok to generate suggestive images of minors and strip clothing from original photos.

Company Responds to Reports

Grok posted on X that isolated cases allowed users to request and receive AI images showing minors in minimal clothing. The chatbot stated that xAI has safeguards but improvements are ongoing to block such requests entirely. In one post, Grok apologized for generating an AI image of two female minors and said the photo violated ethical standards and potentially U.S. law on child pornography.

When CBS News requested comment, xAI responded with the phrase „Legacy Media Lies.“ The company did not provide further details. French officials reported the content to prosecutors on Friday and called it manifestly illegal.

Federal Law and Expert Warnings

Federal law bars the production and distribution of child sexual abuse material. Stefan Turkheimer from RAINN, an anti-sexual violence group, told CBS News that calling these cases isolated minimizes the impact. He said survivors of tech-enabled abuse experience ongoing harm with every notification.

Copyleaks, a content detection tool, found thousands of sexually explicit images created by Grok this week alone. The company warned that AI safety failures are becoming increasingly common. Without strong safeguards and independent detection, manipulated media can be weaponized.

Previous Concerns With Platform

Grok faced similar problems before. The platform launched Spicy Mode last year and framed it as a way for creators to tell edgier narratives. However, a news writer for The Verge tested the technology in August and said the AI model generated unprompted nude deepfakes of Taylor Swift.

Alon Yamin, CEO of Copyleaks, said the impact can be immediate and deeply personal when AI systems allow manipulation of real people’s images without clear consent. One user posted side-by-side photos showing how Grok altered her dress photo into a bikini image. She asked how this was not illegal.

Total
0
Shares
Previous Post
A tightly knotted stethoscope encircling the official Google multicolor G logo like a trap, set on a glossy white table with scattered medical charts as abstract shapes (no legible text), cool blue tones contrasted with urgent red highlights, medium close-up

Google’s AI gives wrong health advice that could hurt people

Next Post
Neutral editorial collage with a faceless engineer silhouette at center holding a glowing laptop, left side a clean Google-colored geometric workspace and right side a clean Anthropic-themed abstract workspace, a large luminous hourglass in the foreground morphing into a fast-spinning clock, behind it a web of small glowing agent nodes connected by lines across a dark-to-bright gradient background, high brightness, crisp cutout montage, no text

Google engineer says AI tool built in one hour what took team a year

Related Posts