eternal-flame-ad/threatexchange

ThreatExchange is a collection of tools that helps platforms detect and remove harmful content like known abuse images or pirated videos. Instead of each company building its own detection system from scratch, these tools let safety teams generate digital "fingerprints" of bad content and share them with each other, so a photo flagged on one service can be automatically caught on another. At its core, the project provides hashing algorithms that convert images and videos into compact signatures. PDQ handles photos, producing 256-bit identifiers. TMK and vPDQ tackle video, each using different approaches to match similar clips. Once content is hashed, tools like the Hasher-Matcher-Actioner platform let teams maintain databases of known bad content and scan new uploads against those lists. A Python library ties it together with reference implementations for downloading shared hashes and running scans. Trust and safety teams at social platforms, messaging apps, or file-sharing services would use this. For example, if a platform wants to block known non-consensual intimate images, it can receive shared hashes from partners, then automatically flag or remove any matching uploads. The tools also support exchanging threat intelligence like malware indicators or phishing URLs through Meta's ThreatExchange API. The project began as code for Meta's API but has grown into a broader toolkit. Notably, HMA is built for AWS deployment, while a newer version called Open Media Match is being designed to work across different cloud providers. Several components are at different stages of completeness, and accessing the full ThreatExchange dataset requires applying for API access through Meta.