Meta AI detector fails to spot its own cropped images

An analysis report released by Reuters claims that Meta's newly previewed AI verification system, routinely fails to identify its own AI-generated images if they have been cropped.

Meta recently showcased a web-based detection tool alongside the launch of its newest image-generation model, "Muse Image."

On its website, Meta says the preview detection tool can identify its own AI-generated images, even if they are cropped, through an invisible watermarking system called " Content Seal" which is embedded in every image generated by Muse Image and designed to help users verify whether it was created by Meta's AI models.

The company explicitly claimed it would remain detectable even if an image was cropped, compressed, or screenshotted.

In an analysis of 40 images generated using Muse Image, Reuters found the detection tool verified all of the original AI-generated images but failed to verify 55% of the same images after they were cropped to approximately one-third to one-half of their original size.

When asked about the results of the Reuters analysis of the detection tool, Meta emphasized that the detection tool is still in a "preview" phase.

Mark Zuckerbergs's company also clarified that while "Content Seal" is engineered to withstand routine, minor digital edits, the embedded tracking signal can be lost if an image is subjected to "heavy cropping."

In March, Meta's oversight board, a body of experts that makes binding decisions and issues recommendations on content issues across the company's social media platforms, called on the company.

Additionally, rival tech companies like Google and OpenAI have also cautioned that their own detection tools are not foolproof against image-alteration techniques.