Is It AI? How to Tell Using Metadata (photoinvestigator.co)

A recent discussion highlights the growing challenge of discerning AI-generated images from real photographs, showcasing incidents where misinformation caused significant real-world effects, such as a false Pentagon explosion image that briefly impacted the stock market. These events underscore the urgent need for effective detection methods in a landscape where viral misinformation spreads rapidly. Notably, Eliot Higgins, founder of Bellingcat, demonstrated the thin line between authenticity and deception by creating AI-generated images that were quickly misinterpreted as real news. To combat this issue, advancements in metadata technology are being utilized. AI image generators like OpenAI's DALL-E 3 and Adobe Firefly now embed content credentials using C2PA manifests, allowing users to verify an image's origin and authenticity. This metadata, along with XMP DigitalSourceType fields and EXIF Software signatures, can provide crucial hints as to whether an image was generated or captured. Tools like the Photo Investigator empower users to perform quick checks on their devices, raising the bar for producing credible deep fakes in the face of misinformation, thus making it more challenging for bad actors to spread misleading content.