Farad

About

Farad is a defensive cybersecurity platform built around simulation. Today it trains and evaluates cooperative multi-agent reinforcement learning defenders against adversaries scripted over a modeled network.

Five specialized agents cover detection, response, hardening, forensics and threat hunting. Benchmark reports reference MITRE-aligned threats and public references such as CAGE-4, and a replayable spatial view shows what a defender did and why.

The longer-term aim is collaborative defense through federated learning and zero-knowledge proofs, so organizations can share threat intelligence while keeping their infrastructure private. The site states that the proof circuits are implemented and that the cross-organization exchange is on the roadmap.

The public demo uses a trained detector to read the stage and severity of synthetic attack scenarios as a fast estimate.

Butts Studios designed the landing site.