Microsoft is finally clamping down on scam attacks with new "scareware" sensor - here's what you need to know (www.techradar.com)

Microsoft is rolling out an AI-driven “Scareware blocker” in Edge (version 142) alongside a new scareware sensor to stop tech-support and other scam pop-ups before users interact with them. The blocker—enabled by default on machines with at least 2 GB RAM and four CPU cores—runs a local computer-vision model to detect deceptive full-screen pages (fake virus alerts, counterfeit control panels, bogus blue screens, and fraudulent law-enforcement payment demands) and automatically shuts them down. Community reports feed Defender SmartScreen’s global blocklist; in initial tests a single user report prevented roughly 50 other targets. IT admins can control the feature via enterprise policies and allow-lists. Technically, the scareware sensor alerts SmartScreen in real time about suspected scams without uploading screenshots or extra user data, and Microsoft says it has upgraded the reporting pipeline to shorten response latency from hours or days to much faster protection. The combination of local inference (privacy-friendly, low-latency detection), centralized threat intelligence, and community reporting creates a rapid feedback loop that can block novel social-engineering attacks before they spread widely. Implications for the AI/ML community include a practical deployment example of on-device computer vision + cloud intelligence, a case study in operationalizing low-latency threat feedback, and an ongoing arms race with adversaries that may push tactics to evade visual models.