Deepfake Protection combines two functions. The first is detection: infringers operate in a deliberately volatile environment engineered to evade it —rotating creatives, disposable ad accounts, short-lived campaigns— and in video-native formats (reels, shorts) where titles, captions, and hashtags carry little or no exploitable text signal. Flagging an incident the moment it surfaces is therefore an adversarial problem: the metadata traditional monitoring relies on is thinning out. Deepfake Protection overcomes this with layered detection. On top of keyword and reverse-image search, it deploys honeypot discovery agents that match the exact victim profile each fraud campaign targets by geography, age, and interest signals. By feeding those signals into each platform’s recommendation and ad-delivery systems, the agents cause the fraudulent organic posts and paid ads to be served to them directly. 

Source: https://www.tennessean.com/press-release/story/209621/red-points-launches-deepfake-protection-to-detect-and-take-down-ai-generated-videos-impersonating-public-figures/