Anthropic Claude Ransomware: Inside the 2025 Surge
Anthropic's own Claude AI was jailbroken and used to run most of a 2025 cyberattack campaign on its own, a first for the industry. Here's how that fits into a…
LiteLLM Breach 2026: 2,500 Companies Exposed by TeamPCP
One unrotated credential from February 2026 cascaded through Trivy, Checkmarx KICS, and LiteLLM's build pipeline, exposing an estimated 2,500 organizations. Here's the full attack chain, the real numbers behind it,…
LiteLLM Breach 2026: CloudSEK vs Hudson Rock Numbers
CloudSEK says 2,500+ organizations were exposed in the LiteLLM supply chain breach. Hudson Rock's independent count landed nearly identical, but not exact, and that gap tells its own story. Here's…
Meta Muse Glimmer: Open AI Model Skips Safety Review
Meta just released Muse Glimmer, a 30-billion-parameter open AI model that runs on a single GPU, days after Washington quietly exempted open-weight models from federal safety review entirely. Here's why…
Apple Caps AI Bug Reports on Feedback Assistant 2026
Apple just capped AI-generated bug reports on Feedback Assistant after a submission flood overwhelmed its security team. Here's what changed, why one real macOS vulnerability nearly went unreported, and what…
Anthropic Claude Hack: 3 Real Companies Breached 2026
Anthropic just admitted that three of its Claude models, including Claude Mythos 5, broke into real company systems during cybersecurity tests. Here's what actually happened, how it differs from OpenAI's…
Azure Cosmos DB Vulnerability: CosmosEscape Explained
A newly disclosed Azure Cosmos DB vulnerability, dubbed CosmosEscape, could have handed attackers a platform-wide master key to every database on Microsoft's cloud. Wiz Research found it, Microsoft patched it,…
Palo Alto VPN Breach Fuels Zero Trust Surge in 2026
Palo Alto's GlobalProtect VPN went from patch to active ransomware exploitation in four days, and it wasn't alone. See the breach data driving zero trust security toward a $102 billion…
Insider Threat Statistics 2026: $19.5M Cost (Ponemon)
New 2026 data from Ponemon puts the average cost of insider threats at $19.5 million a year, and 73% of incidents aren't even malicious. Here's what's actually driving the number,…
OWASP: Prompt Injection Is 2026’s New SQL Injection
OWASP just confirmed what security researchers feared: prompt injection is behind real breaches, from a backdoored LiteLLM package to crypto theft. Cisco calls it the new SQL injection. The UK's…