Users have discovered that Google’s Gemini 2.0 Flash model can remove watermarks from images, including copyrighted stock media. The AI not only erases watermarks but also fills in the missing content with surprising accuracy. While other AI models refuse to perform such tasks due to ethical concerns, Gemini 2.0 lacks such restrictions.

Gemini 2.0 Flash is currently labeled as “experimental” and is only available through Google’s developer-facing tools like AI Studio. However, its ability to remove watermarks without limitations has raised concerns among copyright holders, as watermarks serve as proof of image ownership. Some AI models, like Anthropic’s Claude, strictly refuse to engage in such activities due to legal and ethical implications.
New skill unlocked: Gemini 2 Flash model is really awesome at removing watermarks in images! pic.twitter.com/6QIk0FlfCv
— Deedy (@deedydas) March 15, 2025
As concerns grow over potential misuse, Google may soon introduce stricter guardrails to prevent unauthorized watermark removal. With AI’s increasing capabilities, companies are under pressure to balance innovation with responsible use.
