Heboh! Gemini Latest Version Can Eliminate Image Watermarks

Gemini can delete watermarks (photo: dock. Google)

JAKARTA Gemini became a hot topic of conversation because the big language model became more sophisticated after it was updated. However, this sophistication also caused controversy because it was detrimental to copyright.

Recently, Google launched Gemini 2.0 Flash with a number of upgrades, one of which is the support for more original image creation. This AI model can also edit image content and make the results feel more original.

However, this sophistication has no boundaries like those exhibited by its users on social media. Many users show that Gemini can not only edit images, but can also eliminate watermarks.

This ability of course harms many parties with copyrighted content, ranging from celebrities, media, even to companies. One user tries to test this ability by taking one of the Getty Images pictures.

The medium is known for its large enough watermark in the lower left corner of the image. However, when the image is given to Gemini with the order 'delete the watermark from this image', Gemini will remove the watermark.

What caught my attention was Gemini's watermark. The AI model will automatically add watermarks to the edits. This is a concern because Gemini can remove image watermarks and replace them with their watermarks.

This ability has the potential to be misused because not everyone likes to see watermarks in an image, even though the goal is to show ownership. If freedom of eliminating in this watermark remains available, Gemini will get harsh criticism.

The AI model should provide restrictions regarding the use of its image fabrication or editing tools. GPT-4o and Anthropic version of Anthropic Claude 3.7 Sonnet versions have similar features, but they will reject requests to remove watermarks from any images.