Google’s Gemini AI model is facing criticism for reportedly removing watermarks from images, even as the company works on enhancing transparency in AI-generated content.
According to reports from X (formerly Twitter) and Reddit, users have found that Gemini 2.0 Flash can erase watermarks from stock images, including those from Getty Images, with greater efficiency than other AI tools. A report by TechCrunch highlights concerns over the model’s ability to bypass copyright protections.
In addition to watermark removal, the AI’s text-to-image feature reportedly has fewer safeguards, allowing it to generate images of celebrities and other copyrighted content. This lack of restrictions could lead to legal disputes from copyright holders.
Gemini 2.0 Flash’s image generation feature was recently made available for “experimental” use through Google AI Studio, a platform designed for AI developers. However, it is not yet intended for production use.
Meanwhile, Google DeepMind has been working on SynthID, a tool for watermarking AI-generated content. In 2024, the company made SynthID Text available to the public and recently announced plans to integrate SynthID into Google Photos to mark AI-altered images.
Google is also a key member of the Coalition for Content Provenance and Authenticity (C2PA), alongside tech giants like Microsoft, Amazon, Adobe, and OpenAI. The coalition aims to establish industry standards for tracking AI-generated content using metadata. However, challenges remain in ensuring widespread adoption and interoperability.
