Watermarks on the way: ChatGPT testing adding watermarks to AI images for free users
The reference was spotted in ChatGPT's Android application by a researcher. OpenAI has not yet made any official announcement about this.
- Apr 10, 2025,
- Updated Apr 10, 2025 1:26 PM IST
OpenAI appears to be testing a watermarking feature for images generated by free users of ChatGPT, according to findings in a recent beta version of the Android app. AI researcher Tibor Blaho spotted a reference labelled “image-gen-watermark-for-free” in version 1.2025.091, suggesting that watermarks might soon be applied specifically to visuals created on the free tier.
The image generation tool, which launched as part of the ChatGPT-4o update in early April, has quickly gained popularity, especially for stylised visuals such as Studio Ghibli-inspired art. However, there were also concerns about the image generation tool being used for activities such as creating fake Aadhaar cards. If the watermarking system is implemented, it could be a way to curb the spread of visual misinformation.
The watermarking system is currently in testing and has not been officially announced by OpenAI. Plans may still change before any full rollout, and it's unclear how intrusive or prominent the watermark would be. Free users may still be able to generate such images, but only download versions that carry a visible watermark.
If implemented, this move could have a noticeable impact on how free users share AI-generated content. It appears to be a response to growing concerns about authenticity, copyright confusion, and the increasing difficulty in distinguishing real from synthetic visuals online.
Paid users, such as those subscribed to ChatGPT Plus, are unlikely to be affected based on the current leak, although OpenAI has not made any formal comments on the matter.
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine
OpenAI appears to be testing a watermarking feature for images generated by free users of ChatGPT, according to findings in a recent beta version of the Android app. AI researcher Tibor Blaho spotted a reference labelled “image-gen-watermark-for-free” in version 1.2025.091, suggesting that watermarks might soon be applied specifically to visuals created on the free tier.
The image generation tool, which launched as part of the ChatGPT-4o update in early April, has quickly gained popularity, especially for stylised visuals such as Studio Ghibli-inspired art. However, there were also concerns about the image generation tool being used for activities such as creating fake Aadhaar cards. If the watermarking system is implemented, it could be a way to curb the spread of visual misinformation.
The watermarking system is currently in testing and has not been officially announced by OpenAI. Plans may still change before any full rollout, and it's unclear how intrusive or prominent the watermark would be. Free users may still be able to generate such images, but only download versions that carry a visible watermark.
If implemented, this move could have a noticeable impact on how free users share AI-generated content. It appears to be a response to growing concerns about authenticity, copyright confusion, and the increasing difficulty in distinguishing real from synthetic visuals online.
Paid users, such as those subscribed to ChatGPT Plus, are unlikely to be affected based on the current leak, although OpenAI has not made any formal comments on the matter.
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine