AI & software security
-
Rank Math plugin compromise exposes WordPress sites
Administrators who installed Rank Math PRO during a short October distribution compromise have been warned that upgrading the plugin alone may not remove malicious persistence.
-
Siemens confirms exploitation of Software Center flaw
Siemens has confirmed active abuse of a DLL side-loading vulnerability in Software Center, warning that attackers are distributing a genuine signed executable alongside malicious files.
-
Google brings Wiz runtime detection into GKE with managed preview add-on
Google Cloud has introduced a preview Wiz Runtime Sensor add-on for GKE, offering managed deployment of runtime monitoring across Standard and Autopilot clusters.
-
Microsoft and NVIDIA set out open-weight AI model risk assessment framework
Microsoft and NVIDIA have published guidance on assessing open-weight AI models in enterprise use, separating model-level evaluation from security controls around deployed applications and agents.
-
CERT Polska discloses Windows service control flaw in NordVPN
CERT Polska says a local permissions flaw in NordVPN for Windows before version 8.12 could let low-privilege attackers stop unprotected Windows services.
-
AOMEI backup software campaign used brief update-channel tampering to deploy backdoors
ManageEngine has detailed a supply chain intrusion in which attackers briefly altered AOMEI Backupper configuration to deliver malicious software through a trusted application channel.
-
NGINX patches five conditional web-server security vulnerabilities
New NGINX stable and mainline releases address memory corruption, HTTP/3 request handling and TLS session weaknesses, with each issue requiring a particular configuration or feature.
-
Aoki infostealer expands credential theft to AI coding assistants
Symantec researchers identify recent Aoki samples collecting AI coding assistant credentials. Separately supplied Black Duck commentary outlines potential downstream risks, not confirmed compromises.
-
Palo Alto Networks confirms limited attacks against PAN-OS management flaw
Palo Alto Networks says an authenticated management-interface flaw can give an attacker root-level execution and that very limited attempts have been observed.
-
Drupal fixes critical Field UI flaw with unauthenticated exploit path
Drupal has released security updates for a critical core Field UI vulnerability, explaining that contributed modules and custom code can make a remote execution path reachable without login.