By 
OpenAI is introducing watermarks for images generated by free users of its ChatGPT AI chatbot. The watermarks, which are currently being tested, will be added to images created using the free version of ChatGPT, but not to those generated by paid users. This move aims to distinguish between human-created and AI-generated content, promoting transparency and authenticity.
The introduction of watermarks follows a surge in users creating images using ChatGPT’s ImageGen model, including Studio Ghibli-style images. OpenAI might be trying to encourage free users to upgrade to paid tiers, as paid users won’t have their images watermarked.
The watermarking system is designed to be subtle, avoiding distractions from the image itself. It may use techniques like pixel tweaking to create a pattern detectable by watermark detection tools, similar to Google’s approach.
Other AI companies, such as Gemini and Meta AI, already add watermarks to all their generated images. OpenAI’s decision to watermark only free user images sets it apart. The company has also updated its terms of service, allowing it to reproduce, distribute, and modify content for promotional purposes.
Comments