In a move that could change the way AI-generated content is identified online, OpenAI is reportedly testing watermarks for images created by free-tier users of ChatGPT. The development comes shortly after the company rolled out advanced image generation capabilities of its latest model, ChatGPT-4o, to free users.
Earlier this month, OpenAI unlocked its powerful ImageGen tool—previously exclusive to paid users—for everyone, sparking a surge of interest as users flooded the internet with AI-generated art, including popular Studio Ghibli-inspired creations. Now, according to AI researcher Tibot Blaho, a new line of code discovered in the Android beta version of ChatGPT reads: “image-gen-watermark-for-free”, hinting at upcoming watermarking for free users.
Watermarking AI-generated content has been a long-standing demand from experts, aimed at distinguishing between human-made and machine-generated media. Competing platforms like Google’s Gemini and Meta AI already include visible or hidden watermarks in all their generated images. OpenAI’s apparent decision to watermark only free-tier images marks a different approach—one that may serve both transparency and business goals.
Industry insiders suggest that the move could be a strategic push to encourage more users to subscribe to ChatGPT Plus or other premium plans, which may continue offering watermark-free outputs.
OpenAI has not yet made an official announcement about the watermarking feature. However, speculation continues to rise, especially as the company recently updated its Terms of Service on April 4. The new terms state that any media—images or videos—shared publicly via ChatGPT or on Sora’s explore page may be used by OpenAI to “reproduce, distribute, modify, display and perform it for the purpose of operating and promoting the Services.”
This update further indicates OpenAI’s intention to establish more control over how user-generated AI content is distributed and used online.
As AI tools become more accessible and powerful, questions around authenticity, misuse, and accountability are also rising. Watermarking is seen as a step towards responsible AI usage, helping platforms, creators, and audiences verify the origin of content.
For now, free-tier users of ChatGPT can continue to experiment with ImageGen without visible watermarks, but that may soon change. Whether this feature will also be applied retroactively or just to future generations remains to be seen.
