Google’s Gemini 2.0 Flash reportedly possesses the capability to remove watermarks and has also demonstrated the ability to generate AI-generated celebrity photographs.

Gemini 2.0 Flash: Watermark Removal and Celebrity Image Generation Capabilities

• Google’s Gemini 2.0 Flash AI model has reportedly demonstrated the ability to remove watermarks from images and generate images of celebrities, raising concerns about copyright infringement and potential legal issues.

• Reddit users have showcased Gemini’s capacity to remove watermarks from copyrighted stock images, potentially violating DMCA provisions by altering copyright management information, and to create images of celebrities, which may infringe on their rights of publicity.

• While Google labels the image generation feature as experimental and available only in Google AI Studio,  the inconsistencies in watermark removal functionality (successful for some users, unsuccessful for others) highlight the need for further refinement and stricter controls.

• The capabilities of Gemini 2.0 Flash underscore the ongoing ethical and legal challenges surrounding AI image generation, emphasizing the importance of responsible development and implementation of AI technologies to prevent misuse and copyright infringement.

STORY

Google’s Gemini 2.0 Flash model has generated controversy due to its ability to reportedly remove watermarks from user-uploaded images and create images of copyrighted characters and celebrities. Users on Reddit have claimed that the model can eliminate watermarks from licensed stock images—particularly those from major sources like Getty Images—thus raising concerns over potential litigation from media licensing firms. Although preliminary tests indicated that watermark removal was ineffective, the generation of celebrity images was successful.

The model was initially introduced in December for testing and was recently made accessible worldwide at no cost, featuring various functionalities, such as storytelling through pictures, conversational image editing, and improved world understanding and text rendering. AI image generators like Gemini have encountered ongoing ethical scrutiny, leading companies such as Google, Anthropic, and OpenAI to implement strict guidelines against the misuse of their technologies. These guidelines typically prohibit actions like generating images of celebrities and watermark removal from copyrighted material.

Despite these prohibitions, many Reddit users reported that they were able to use Gemini’s image output capabilities to remove watermarks and generate celebrity images. Removing watermarks potentially violates several provisions of the Digital Millennium Copyright Act (DMCA) by altering the Copyright Management Information (CMI) attached to copyrighted works. The legality of creating images of celebrities with AI is complex due to concerns regarding rights of publicity, deepfake technology, and potential defamation.

While some users achieved successful watermark removal, further investigation by the original reporting revealed failed attempts with Gemini’s model, which stated its inability to directly edit images. Nevertheless, conversational image editing did function well, allowing the reproduction of celebrity images, indicating that the tool is still undergoing refinement.

Google has categorized the image generation aspect of Gemini 2.0 Flash as experimental. This status might explain its limited availability in Google AI Studio, as opposed to Gemini’s main site, and could account for inconsistencies in performance. The company’s adoption of an experimental label suggests evolving capabilities and potential adjustments based on user interaction and feedback.

Concerns persist regarding the unauthorized use of altered images, especially for commercial purposes, given the likelihood of facing multiple DMCA claims. While individual users may seek to eliminate watermarks for personal use, the legal implications remain significant. Overall, these developments highlight the ongoing challenges and potential risks associated with AI image generation capabilities, particularly as they pertain to copyright law and ethical standards in media.