Deepfake detection tries to identify whether video, images, or audio have been generated or manipulated with AI. As face-swap and voice tools have become easy to use, people can create convincing fake content of creators without their consent, causing serious reputational and financial harm.
Detection approaches look for inconsistencies that generation tends to leave behind, such as flicker in lighting or reflections, lip-sync mismatches, and unusual patterns in fine image detail. Provenance standards such as C2PA work from the other direction, by attaching verifiable information to genuine content. No method is perfect, and detection gets harder on compressed or re-uploaded video.
Privly can help find content that uses your likeness and pursue removal through the appropriate routes, which for deepfakes may include platform reporting, Google's tools for removing non-consensual explicit imagery, and, where your own photos or footage were used, copyright notices. If you're dealing with a deepfake right now, Privly's free NCII support program can help even if you're not a customer.