OpenAI’s GPT-4o model has recently introduced image generation and photo restyling capabilities, although it may soon implement watermarks for users on its free tier. Watermarking AI-generated media has become an important topic in the industry. Currently, OpenAI is testing this feature specifically for free-tier users.
The introduction of watermarks is believed to encourage users to upgrade to premium subscriptions, providing an incentive to explore paid features. Since its public debut, OpenAI has concentrated on enhancing the user experience of ChatGPT. Initially, the model did not offer image generation, as that function was largely managed by specialized tools like Midjourney.
However, advancements have blurred the lines, as now chatbots, including those from competitors like Meta AI and Google’s Gemini, have begun to provide similar capabilities. In contrast, OpenAI’s GPT-4o model was relatively late to the image generation arena. Released last week for free access after a brief subscriber-only period, GPT-4o allows users to generate images based on textual prompts or restyle existing images.
The model has sparked excitement, especially with its ability to recreate artwork in styles such as that of Studio Ghibli. To manage the high demand, users on the free tier are likely limited to generating only two images per day. Despite its popularity, the lack of watermarking has raised concerns.
AI researcher Tibor Blaho highlighted that a recent version of the ChatGPT Android app refers to a forthcoming ImageGen watermark. This watermark is set to apply only to the outputs created by free-tier users. OpenAI’s selective approach seems unusual, resembling practices of stock image platforms that impose watermarks on free downloads to promote paid options.
As the industry pushes for clear distinctions between AI and human-generated content, it remains to be seen how users will respond to this development once implemented.
Leave a Reply