Photos posted on Facebook and Instagram are marked as being taken by AI. Meta introduced the watermark to prevent the spread of fake images.

Several photographers complain that their photos shared on Facebook and Instagram were tagged “made by AI” by Meta. Meta uses algorithms to distinguish real photos from images created by AI image generators. The algorithms scan the metadata of the image file.

Meta tags

OpenAI and Google, among others, have developed such meta tags that are invisible to the human eye but can be recognized by machines. The photos that now receive the AI ​​watermark may have been enhanced using artificial means.

The algorithms take their job very seriously and therefore equate even the smallest artificial editing with the complete generation of an image by AI. Photographers think this is too far and also find it unclear how exactly Meta’s algorithms work. They no longer receive the label for their work once Meta has awarded it. Others are of the opinion that editing photos is also a distortion of reality and that the term is therefore not wrong.

Real or fake

A Meta spokesperson told TechCrunch and The Verge that the company is aware of the photographers’ complaints and is working with AI watermarking developers to figure out how to further refine the technology. Meta launched AI watermarking in February to alert users about fake images circulating on its platforms.

Such tools are necessary, but this incident shows that the line between real and fake is very thin and that algorithms do not always know this well.