A Powerful and Controversial Functionality
Watermark protection is a common practice among image banks and creators seeking to safeguard their copyright. However, Google’s new artificial intelligence model, Gemini 2.0 Flash, has demonstrated the ability to remove these watermarks in some cases, raising serious concerns about intellectual property and the security of visual content in the era of generative AI.
Google Distances Itself: “It’s a Violation of Our Terms”
Google has condemned the use of this technology to remove watermarks, making it clear that doing so represents “a violation of our terms of service.” Nevertheless, experiments conducted by users and experts have shown that AI can accomplish this with a certain degree of success, generating concern among image banks, designers, and content creators.
How Gemini 2.0 Flash Works in Image Generation
The Gemini 2.0 Flash family of models has surprised in various applications, standing out for its speed and multimodal capabilities. From Google AI Studio, users can access the tool by selecting “Gemini 2.0 Flash (Image Generation) Experimental” in the “Create Prompt” section. This advanced version not only allows generating images from text but also modifying existing images, for example, changing clothing colors or adjusting specific elements within a photograph.
AI and Watermark Removal
Several users have shared their tests with this technology on social media, showing examples where watermarks almost completely disappear. However, the process isn’t always perfect: in some cases, the AI generates distortions or alters visual elements unpredictably. In one of our tests, we tried to remove a previously generated watermark, but instead of doing so precisely, the AI modified the model’s face and removed other details from the image.
Still, when it works correctly, the results are surprising, which opens a new debate about the impact of artificial intelligence on copyright protection.
A New Challenge for Intellectual Property
The advancement of generative AI has already generated controversy in the copyright realm. Recently, OpenAI asked the U.S. government to eliminate restrictions on the use of copyrighted works in AI models, arguing that these barriers limit technological development. Now, with the ability to remove watermarks, the discussion intensifies.
Image banks depend on these protections to ensure creators receive income for their works. If technology advances to the point of making watermarks ineffective, it could put the business model of these platforms at risk and affect thousands of artists and photographers.
Conclusion: Ethics and Regulation in AI
While Google’s technology wasn’t specifically designed to remove watermarks, its ability to do so poses an ethical and legal dilemma. Companies and legislators will need to address these challenges to ensure a balance between technological advancement and copyright protection.
The debate is served: should the use of generative AI in image editing be more strictly regulated? What measures should content creators take to protect their work in this new digital landscape?
