Anthropic's large language model, Claude, identifies previously unknown vulnerabilities in cryptographic protocols, demonstrating AI's potential in security research.

A seemingly simple 'burn-after-read' feature is undone by chat apps unfurling link previews, consuming the single read.

Attackers register non-existent AI-generated package names to ensnare developers.

Thousands of internet-connected servers are exposing sensitive password hashes due to a long-standing vulnerability in their BMC interfaces.

Despite DMARC's public availability since 2012, the vast majority of domains lack critical email authentication, leaving them vulnerable to spoofing.

New API offers real-time web page sanitization and prompt injection detection for AI agents.

A critical licensing service failure outside Xbox's direct control rendered games unplayable, affecting consoles across three generations.

Google's new enterprise security initiative, 'Beyond Zero,' addresses the evolving threat landscape driven by AI agents and new development tools.

OCI is implementing a Zero Trust model for database access, moving beyond traditional perimeter security.

Apple releases critical security patch for macOS Tahoe, fixing flaws in WebKit, Kernel, and more.