theverge[.]com/2024/6/24/24184795/meta-instagram-incorrect-made-by-ai-photo-labels
With Aiarty on GOTD, it's maybe a good time to point out an issue that's starting to drive photographers nuts. Everyone, including governments, is pressuring big companies in media, social media, and software to do something about AI generated fake images, voices, and videos, which can all be used in misinformation campaigns and/or fraud. So far no one has figured out any potential solutions, but that doesn't stop big biz from feeling that they have to do something, if only to somehow give them a little bit of a legal shield. One result of that is embedded tags in image files if AI was used. The software of course has no idea how or why AI was involved, whether it was used to substantially alter a photo, generate the majority of the photo itself, or fix a 3 pixel speck of dust. All 3 can qualify for the tag, meaning those photos will now be branded as AI generated. Not every app will add those AI tags, and not every site or viewer will look for them. A Google search didn't come up with a way to remove AI related tags, other than a trick to skip the tag by copy /pasting a layer into a new photo in Photoshop. Those generating images and such for illicit use apparently don't use software that adds the tags, or so far don't care, which could be taken to mean that current efforts tagging photos as AI are only penalizing legitimate use by regular folks.