Our AI Security Verdict reads web pages written by the people we are trying to catch. We attacked it ourselves and found that one hidden paragraph could talk it out of a phishing verdict. This is what prompt injection is, how it affected us, and how we fixed it.
Jev from TypeSafe AI made typed-decision models the hottest thing in AI this month. We benchmarked 17 open Jev-style models on real ScanMalware verdicts to find one that can double-check our AI Security Verdict. An off-the-shelf model with no training won, two models now check live verdicts, and our first benchmark taught us more by being wrong.
How ScanMalware spots brand impersonation and makes the evidence easier to explore. Browse brand pages, download detections, or flag an incorrect listing.
Deep technical analysis of ShinyHunters' sophisticated real-time phishing kit used for SSO credential theft and MFA bypass. Detailed breakdown of the attack infrastructure, JavaScript malware, and MFA bypass mechanisms used against healthcare and fintech companies.