Multiple vulnerabilities in PHP have been disclosed, allowing attackers to exploit use-after-free conditions leading to remote code execution and SQL injection. Specifically, CVE-2026-6722 and CVE-2026-7261 relate to…
Multi-agent AI systems are vulnerable to privilege abuse when agents delegate tasks across complex chains. This risk, classified by OWASP as Identity & Privilege Abuse, can lead to agents exceeding their original…