People are using Google’s new AI model to remove watermarks from images

On social media, users have discovered a controversial case for Google’s new Gemini AI model: removal of watermarks from photos, including photos published by Gati Images and other well -known stock media organizations.

Last week, Google increased its access Gemini 2.0 flash The model’s image generation feature, which allows the model locally to produce and edit image content. It’s a Powerful CapabilityThrough all accounts. But there are some guards in it. Gemini 2.0 Flash will produce images of abnormally To reflect celebrities And The characters of the right postAnd – as mentioned earlier – remove water marks from existing images.

As several x and reddit Consumers noted, Gemini 2.0 flash will not only remove water marks, but tries to fill any gaps created by the delete of watermark. Other AI-powered tools also do this, but Gemini 2.0 flash seems to be extraordinary in it-and is free to use.

Clearly, the Gemini 2.0 Flash Image Generation feature is currently labeled as “experimental” and “not not Not Not Not Not”, and it is only available in Google’s developer tools. To study. The model is also not a perfect watermark removal. Gemini 2.0 flash struggles with some semi -transparent water marks and water marks that canvas large parts of the images.

Nevertheless, some copyright holders will certainly take up the issue with a lack of lack of restrictions on the use of Gemini 2.0 flash. Model including anthropic Claude 3.7 Swant And Openai’s GPT-4O Clearly deny removing water marks. Claude said, removing a watermark from a photo “immoral and potentially illegal.”

Removing the watermark without the original owner’s consent is considered illegal under US Right to Publication (according to Laws like thisJes Out of unusual exception.

Google did not immediately respond to the comments sent out of normal business hours.