Anthropic Report Details AI Misuse by Yemen Weapons Cell
Bruce Schneier flags Anthropic's new misuse report, which identifies a Yemen-based cell using Claude for weapons development.
Breaches, offensive and defensive AI, and surveillance.
Bruce Schneier flags Anthropic's new misuse report, which identifies a Yemen-based cell using Claude for weapons development.
A researcher's latest Windows Defender exploit underscores how security tools themselves can become attack surfaces.
A new essay argues that post-9/11 mass surveillance has failed its promises and should be dismantled in favor of targeted oversight.
A new YARA rule from SANS ISC helps analysts spot hidden VBA projects inside AutoCAD DWG files.
A beached California boat is holding an estimated 30 to 50 tons of decomposing squid, creating a severe cleanup problem.
A Help Net Security weekly roundup ties a $93M wallet drain to a reseller's devices and details an npm Trusted Publishing flaw that exposed 1,500+ packages.
As AI accelerates and scales vulnerability discovery, security teams must adapt their validation processes to keep up.
Bruce Schneier highlights the Harvest, an IBM-built NSA supercomputer from the 1960s, as an early example of specialized cryptanalytic hardware.
A critical flaw in LiteSpeed Web Server Enterprise could let a low-privileged account take over a shared-hosting server, cPanel warns.
A newly disclosed hardware attack, DDRop, undermines Intel TDX and AMD SEV-SNP by silently dropping memory writes, leaving processors to act on stale encrypted data.
Apple's latest cross-platform update fixes a record 261 vulnerabilities, the most ever addressed in a single release.
A threat actor used a legitimate remote-management tool to maintain persistent control inside a major Thai broadband provider's network.
WordPress will automatically screen every plugin release for security risks before it is distributed through the update API.
A rogue browser extension posing as a Twitch viewer tool exposed nearly 31,000 users' OAuth tokens to a Russian commercial bot service.
New analysis warns that attackers can manipulate AI's defensive reasoning to silently compromise target networks, making AI governance urgent.
A new joint advisory signals a regulatory shift toward transparent incident reporting and away from corporate spin.
The Papercut AI swarm attack illustrates how attackers now use AI across every stage of the intrusion lifecycle, from staging to exfiltration.
A new campaign uses Android's Work Profile feature to hide a banking trojan inside cloned apps aimed at Indonesian users.