TrendAI™ is deepening its focus on the CISA KEV Catalog as a proven signal of active risk. With AI, TrendAI™ combines TrendAI™ ZDI research, exploit intelligence, exposure context, and business risk to drive continuous, AI-assisted prioritization, helping security teams make better decisions and act faster.
TrendAI™ Research uncovered an open redirect in Dify's post-login flow that could have handed a freshly authenticated session, token and all, to an attacker, and worked with the vendor to close it across every sign-in path before the details went public.
The Singapore Police Force (SPF), working closely with Pakistan's National Cyber Crime Investigation Agency (NCCIA) and INTERPOL has arrested two individuals linked to Tycoon2FA, a phishing operation that served criminal customers across four continents.
Disruption reported across at least seven states, from equipment left accessible online. The issue is largely a matter of configuration and access control, and here's what to fix first.
TrendAI™ is deepening its focus on the CISA KEV Catalog as a proven signal of active risk. With AI, TrendAI™ combines TrendAI™ ZDI research, exploit intelligence, exposure context, and business risk to drive continuous, AI-assisted prioritization, helping security teams make better decisions and act faster.
TrendAI™ Research uncovered an open redirect in Dify's post-login flow that could have handed a freshly authenticated session, token and all, to an attacker, and worked with the vendor to close it across every sign-in path before the details went public.
The Singapore Police Force (SPF), working closely with Pakistan's National Cyber Crime Investigation Agency (NCCIA) and INTERPOL has arrested two individuals linked to Tycoon2FA, a phishing operation that served criminal customers across four continents.
Disruption reported across at least seven states, from equipment left accessible online. The issue is largely a matter of configuration and access control, and here's what to fix first.
TrendAI se une a Nvidia como socio inaugural en la Open Secure AI Alliance, avanzando en modelos abiertos, arneses e investigación para fortalecer la defensa cibernética.
Los propios modelos de OpenAI salieron de un entorno de prueba y entraron en los servidores de Hugging Face para resolver una evaluación, sin la intervención de un atacante humano. El incidente demostró que mantener la seguridad de la IA agente ahora depende de cómo se contiene, no solo de cómo se entrena.
Un ataque a la cadena de suministro afectó a Axios cuando los atacantes utilizaron credenciales npm robadas para publicar versiones maliciosas que contenían una dependencia fantasma. Esto activó un RAT multiplataforma durante la instalación y reemplazó sus archivos con señuelos limpios, lo que dificultó su detección.
PeckBirdy es un sofisticado marco de C&C basado en JScript utilizado por grupos APT alineados con China para explotar LOLBins en múltiples entornos, entregando puertas traseras avanzadas para atacar a las industrias de juegos de azar y entidades gubernamentales asiáticas.
CVE-2025-55182 es un RCE de preautenticación con CVSS 10.0 que afecta a los Componentes del Servidor de React. En medio de la avalancha de exploits de prueba de concepto falsos, escáneres, exploits y conceptos erróneos generalizados, este análisis técnico pretende aclarar la confusión.
A tokenizer lies at the core of every large language model. When it drifts, whether from unseen flaws or adversarial interference, costs rise and performance drops. We explore this emerging risk, its implications, and the measures to prevent it.
August 6th, 2026 
5 min read
