OpenAI, the developer of ChatGPT, has reportedly begun testing watermarks for images created with the free version of ChatGPT.
A watermark that appears discreetly on images created by artificial intelligence is designed to provide clear labeling of AI-generated content. This innovation could be a response to concerns about the authenticity and copyright of AI-generated content.
Artificial intelligence researcher Tibor Blaho discovered references to the “ImageGen” watermark in the code of the ChatGPT image generation model. However, the watermark is currently only being tested for free accounts, and it is unknown whether it will be extended to premium subscriptions or third-party APIs.
OpenAI has yet to officially announce the implementation of watermarks or set a timeline for their launch, but competitors such as Meta AI and Google Gemini already use visible watermarks for their AI-generated images.