Google’s latest AI model, Gemini 2.0 Flash, is making headlines for a controversial reason—its ability to remove watermarks from images, including those from stock photo giants like Getty Images. This discovery, made by users on X (formerly Twitter) and Reddit, has raised significant concerns about copyright violations and misuse. While AI-powered watermark removal is not new, Gemini 2.0 Flash appears to be particularly effective and, most importantly, free to use.
A Powerful Editing Tool With Few Restrictions
Last week, Google expanded access to Gemini 2.0 Flash’s image generation and editing capabilities through its developer tools, including AI Studio. This lightweight model allows users to create and modify images conversationally, making it easier than ever to manipulate digital content.
Read More: ChatGPT can now replace Gemini as the default assistant on …
However, users quickly found that Gemini 2.0 Flash could erase watermarks with remarkable accuracy. Unlike other AI-powered watermark removers, this model does more than just delete—it fills in the gaps left behind, reconstructing the image seamlessly.
Notably, some watermarks still pose a challenge, particularly semi-transparent ones or those covering large portions of an image. However, the tool’s ability to handle complex watermarks, such as Getty Images’ prominent stamps, has set it apart from other AI editors.
Google’s AI Replaces Watermarks With Its Own Markings
Interestingly, after removing a watermark, Gemini 2.0 Flash does not leave the image completely unmarked. Instead, it replaces the original watermark with Google’s SynthID mark, a digital fingerprint indicating the image has been altered with AI. But as past examples have shown, AI-generated marks can also be removed using AI. Similar tools, like Samsung’s object erase function, have demonstrated how easily such watermarks can be erased, further complicating the issue.
Ethical and Legal Implications
The discovery of Gemini 2.0 Flash’s watermark removal ability has led to concerns from copyright holders and legal experts. Removing a watermark without the original owner’s consent is illegal under U.S. copyright law, except in rare circumstances.
Competing AI models like OpenAI’s GPT-4o and Anthropic’s Claude 3.7 Sonnet refuse to process requests for watermark removal, citing ethical concerns. Claude even labels such actions as “unethical and potentially illegal”.
Despite this, Google has yet to respond to inquiries regarding the potential misuse of its AI model. The company’s decision to keep Gemini 2.0 Flash’s image generation feature “experimental” and “not for production use” suggests that it may still be working on introducing more guardrails.
Gemini 2.0 Flash Can Also Generate Images of Celebrities
Beyond watermark removal, users have discovered another controversial aspect of Gemini 2.0 Flash—it can generate images of real people, including celebrities like Elon Musk. This capability is particularly surprising because Google’s full Gemini model typically restricts the creation of recognizable individuals. However, the lighter Flash model seems to bypass these protections, adding another layer of ethical and legal concerns.
Read More: Pakistan joins Google Wallet for digital transactions
For now, Gemini 2.0 Flash’s image editing features are only accessible to developers through AI Studio, meaning it isn’t widely available for the general public. But as AI technology advances, the debate over copyright protection, ethical AI use, and content ownership will only intensify. With growing concerns from content creators and stock image companies, Google may soon be forced to address the issue and implement stricter safeguards. Whether or not these changes come in time to prevent widespread misuse remains to be seen.