X has introduced a new setting on its iOS application designed to limit photo manipulation via the Grok artificial intelligence chatbot, though the feature offers only partial protection. According to reports verified by The Verge and Social Media Today, a toggle labeled "block modifications by Grok" was added to the image upload settings. However, the feature does not entirely prevent the AI from editing images. The fine print clarifies that users can only prevent the specific tag @Grok from modifying the content within the social media post. The primary motivation behind this update is to address the abuse of Grok's image editing capabilities, which were recently utilized to generate undressed photographs of real individuals, including children. Previously, the ability to tag the xAI chatbot in replies to edit images was restricted for free accounts following backlash from lawmakers and regulators. This new toggle extends that restriction to Premium subscribers for the specific tagging mechanism, effectively stopping paid users from using that particular method to alter images in the timeline. Locating the feature is not straightforward. On the iOS app, users must tap the paintbrush icon on an image thumbnail within the post builder, then select a flag icon on the editing taskbar. The option does not appear during the upload process on the web version, nor does it apply to photos already posted before the feature was released. Testing confirmed that while the toggle blocks the tagging method, it leaves other avenues for manipulation open. For instance, a user can long-press a protected image on the iOS app to trigger a direct prompt that opens the photo in the Grok application independently. This direct link allows image editing without restriction. Additionally, users can save a protected image, re-upload it to the same thread, and tag Grok again since the new tag does not persist on reposted content. The Verge noted that the toggle fails to protect against these alternative methods, limiting the feature's effectiveness as a comprehensive safety tool. The availability of this function remains uncertain as X has not issued an official announcement regarding its release or long-term status. It is unclear if the feature will expand to the web platform or older content in the future. The current iteration serves as a specific countermeasure to the tagging abuse rather than a total ban on image manipulation. X was contacted for further comment on the scope and future development of the tool. While the update represents a step toward addressing safety concerns, it highlights the difficulty of implementing robust protections in a platform that allows deep integration between social interaction and generative AI tools.