Social media users have discovered a controversial use case for Google’s new Gemini artificial intelligence model: removing watermarks from images, including those published by Getty Images and other well-known stock media resources.
Last week, Google expanded access to the image generation feature in Gemini 2.0 Flash, which allows the model to generate and edit image content independently. By all accounts, this is a powerful feature. But it also seems to have a few limitations. Gemini 2.0 Flash will create images featuring celebrities and copyrighted characters without any complaints, and as mentioned earlier, it will remove watermarks from existing photos.
As several X and Reddit users have pointed out, Gemini 2.0 Flash doesn’t just remove watermarks, it tries to fill in any gaps left by the watermark removal. Other AI tools do this as well, but Gemini 2.0 Flash seems to be extremely adept at it – and it’s free to use.
To be clear, the image generation feature in Gemini 2.0 Flash is currently labeled “experimental” and “not for production use” and is only available in Google’s developer tools like AI Studio. The model also does not remove watermarks perfectly. Gemini 2.0 Flash doesn’t seem to be able to handle some translucent watermarks and watermarks that are overlaid on large parts of images.
However, some copyright holders will probably be unhappy with the lack of restrictions on the use of Gemini 2.0 Flash. Models including Anthropic’s Claude 3.7 Sonnet and OpenAI’s GPT-4o explicitly refuse to remove watermarks; Claude calls removing a watermark from an image “unethical and potentially illegal.”
Removing a watermark without the original owner’s consent is considered illegal under U.S. copyright law (according to law firms like this one), with rare exceptions.
