OpenAI is reportedly developing new features for ChatGPT, including a potential watermarking system for AI-generated images, a development initially noted in April 2025 by Tibor Blaho on X.
According to insights gained from the latest ChatGPT v1.2025.196 beta, the code for a “Save without watermark” option has been revealed. Screenshots indicate that this option would be accessible via a three-dot menu on generated images, alongside existing choices like Edit, Select, Save, and Share. The feature was activated on the free tier during testing, but its final implementation could vary.
The introduction of a “Save without watermark” option suggests several possibilities for the watermarking strategy. One scenario is that all ChatGPT-generated images might carry a watermark, but paid subscribers could be given the exclusive ability to save images without it. Another possibility is that all users, regardless of subscription status, might have the option to remove the watermark, though the process could be designed to introduce enough friction to discourage casual users from doing so.
This potential feature is part of a broader set of new functionalities OpenAI is exploring for ChatGPT, which also includes “Study Together,” “Image Styles,” and a yearly subscription plan. It is essential to note that information derived from APK teardowns, while indicative of features under development, does not guarantee their eventual public release.
OpenAI has not yet officially announced this watermarking feature, meaning there is still a possibility that the concept could be refined or even abandoned. The company’s ongoing development efforts aim to enhance user experience and functionality within ChatGPT.
