Google has released a new version of its Gemini 2.0 Flash AI model that can generate its own images, but the tool raises concerns over its ability to remove watermarks from images. The mobile app and browser-based versions of Gemini use Imagen 3, Google’s flagship AI image generator, which does not remove watermarks. In contrast, Gemini 2.0 Flash accessed via AI Studio, a developer-focused interface, can remove watermarks with surprising accuracy.
This raises concerns about the potential for large-scale copyright infringement, as users have reported easily removing watermarks from images using the tool. Digital watermarks are used to protect digital creators’ work and prevent unauthorized use. Removing these watermarks without permission is considered unethical and potentially illegal.
Additionally, Gemini 2.0 Flash’s ability to add celebrities into images has also been noted, but the results seem inconsistent. The tool’s power of image generation has sparked debate among users, with some finding it useful and others expressing concerns about its potential misuse.
Google has enabled access to this tool through AI Studio, a free interface that requires only a normal Google account for registration. TechRadar has reached out to Google for comment on this issue, but the company has not yet responded. The lack of guardrails around this image generator is concerning and highlights the need for urgent attention from Google to address these issues.
Source: https://www.techradar.com/computing/artificial-intelligence/worryingly-google-geminis-new-ai-image-generation-features-can-be-used-to-remove-watermarks-from-images-and-im-concerned