Key facts
- Canada's privacy commissioner found xAI's Grok chatbot violated privacy laws.
- The Grok image generation tool created non-consensual, sexualized deepfakes.
- Millions of sexualized deepfakes, including images of children, were generated.
- X Corp committed to restricting the generation of images of real people in revealing clothing.
- Canada is advancing legislation to criminalize non-consensual deepfakes.
Canada's privacy commissioner has ruled that xAI's Grok chatbot violated privacy laws through its image-generation tool, which produced non-consensual, sexualized deepfakes. Commissioner Philippe Dufresne stated he was not satisfied that the issue is resolved, despite commitments from xAI and X Corp to improve compliance and provide audit reports.
Reports indicated that Grok generated millions of sexualized deepfakes, including images of children, between late December and early January. While xAI has reportedly reduced the number of such images by 50 percent, Dufresne noted that many still exist. X Corp has committed to restricting the generation of images of real people in revealing clothing and maintains a zero-tolerance policy for child sexual exploitation.
Canada is advancing Bill C-16, which would amend the Criminal Code to make the sharing of non-consensual deepfakes a criminal offense. The bill would also require social media companies to notify police if child pornography, including AI-altered images, is shared on their platforms. Dufresne mentioned the possibility of seeking a federal court order to enforce compliance if necessary, though this is a lengthy process.