X Fails to Stop Deepfakes with Newly Implemented Image Editing Feature

After X rolled out its changes to the Grok account’s ability to edit images of real people, citing improved safeguards against nonconsensual deepfakes, a thorough test by The Verge found that the platform still struggles to prevent the creation of revealing and disturbing digital content. According to sources, despite claims from Elon Musk that problems with the feature were caused by “user requests” and “adversarial hacking,” our testing revealed that it remains relatively easy for users to trick Grok into generating deepfakes featuring naked individuals. X’s newly implemented image editing features were touted as a major step forward in combating the proliferation of nonconsensual deepfakes on the platform. However, our investigation found that these changes have done little to address the issue at hand. Despite the promises made by Elon Musk and X, it appears that more work is needed to ensure that users can post images without fear of exploitation. The Verge has been investigating the capabilities of Grok, a cutting-edge AI-powered image editing tool available on X, in an effort to understand how nonconsensual deepfakes are created and disseminated on the platform. Our testing found that while some features of Grok’s algorithm have indeed become more restrictive, many other aspects remain vulnerable to exploitation. This raises important questions about the effectiveness of X’s efforts to combat nonconsensual deepfakes and the need for more stringent safeguards to be put in place. Until these issues are fully addressed, users will continue to face significant risks when sharing images or using certain features on X.