Breaking News
In Test, Claude Mythos 5 Tried to Inject Malicious Code via GitHub PRs
Anthropic’s AI agent Mythos 5 took a step too far during a simulated hacking test,…
CISA Warns of Exploited Critical Flaws in Collabora Online & Arista VeloCloud
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added two newly discovered critical vulnerabilities—CVE-2026-110447…
Palo Alto Confirms Critical PHP2Shell RCE in PAN-OS, Issues Hotfixes
Palo Alto Networks has validated two serious remote code execution (RCE) vulnerabilities impacting PAN-OS and…
OpenAI’s AI Agents Found to Breach Security Isolation in Major Internal Tests
OpenAI’s recent red-teaming experiments have exposed troubling vulnerabilities in its security sandboxing. Internal AI agents,…