Gemini 2.0 Flash's image generation feature reportedly excels at watermark removal.
The tool’s limited safeguards may enable unauthorized creation of copyrighted content.
Google Photos will implement SynthID to label AI-manipulated images for authenticity.
Google’s Gemini 2.0 Flash AI model has been criticized for its ability to remove watermarks from images. Despite Google’s ongoing efforts to improve AI-generated content, this capability raises concerns about copyright infringement and digital content security. Many users on social media platforms such as X and Reddit claimed that Gemini 2.0 Flash can successfully remove watermarks from images published by stock media websites.
According to TechCrunch, Gemini 2.0 Flash’s free experimental image generation feature outperformed other AI tools in terms of watermark removal accuracy. According to the report, the model’s text-to-image generation feature has fewer restrictions and can generate content depicting celebrities and copyrighted materials.
No way this lasts…
— Min Choi (@minchoi) March 17, 2025
Google Gemini 2.0 Flash can remove watermarks
6 wild examples: pic.twitter.com/lyoRRESlHr
RIP Watermarks
— Poonam Soni (@CodeByPoonam) March 17, 2025
It's only been 4 days since Google dropped Gemini Flash Experimental, and people are going crazy over it's ability to remove image watermarks!
12 wild examples so far: (Don't miss the 5th one) pic.twitter.com/sjERpFN4Vl
New skill unlocked: Gemini 2 Flash model is really awesome at removing watermarks in images! pic.twitter.com/6QIk0FlfCv
— Deedy (@deedydas) March 15, 2025
Holy!!! Gemini Remove Watermarks in 5 sec pic.twitter.com/aiTWT7FBMf
— Jacques Gariepy (@JacquesGariepy) March 17, 2025
Google has not acknowledged the issues at this time, but if true, the lack of strict safeguards and the ability to erase watermarks without consent could result in yet another legal challenge from copyright holders.
It’s worth noting that the Gemini 2.0 Flash image generation capabilities were recently expanded for experimental purposes and are now available through Google AI Studio. However, the company stated that the feature is not meant for production use.
Also read: Scam alert! Retired officer falls victim to Facebook scam, loses Rs 5 crore
In 2024, Google DeepMind introduced SynthID Text, a tool for watermarking AI-generated text. Furthermore, Google Photos will use SynthID to label AI-manipulated images in order to improve content authenticity.
Google is also a member of the Coalition for Content Provenance and Authenticity (C2PA), an industry association that includes Amazon, Microsoft, OpenAI, Adobe, and others. The C2PA has also created technical standards to track the origin of AI-generated images by embedding metadata with information such as the creation date and AI tools used.
Even after these initiatives, the industry continues to face significant challenges in terms of widespread adoption and seamless integration of these standards.
Follow Us
Ashish Singh
Ashish Singh is the Chief Copy Editor at Digit. Previously, he worked as a Senior Sub-Editor with Jagran English from 2022, and has been a journalist since 2020, with experience at Times Internet. Ashish specializes in Technology. In his free time, you can find him exploring new gadgets, gaming, and discovering new places. View Full Profile
