A critical authentication bypass in a widely used LLM proxy shows how a classic web flaw can become far more serious when it hits the control plane.
A reported LiteLLM flaw chain shows how a proxy that concentrates access, secrets, and admin power can turn a low-privilege account into a gateway-level security event.
A LiteLLM vulnerability chain underscores how one command-injection path and one Host-header trust flaw can collide into a high-risk control-plane exposure.
A high-severity LiteLLM command-injection bug shows how AI gateway control panels can collapse into host-level risk when role checks are too loose.
Palo Alto Networks’ Portkey acquisition shows how control over prompts, routing, and agent behavior is turning into the new security battleground.
A reported campaign tied to TeamPCP shows how a single AI middleware package can become a high-value path to secrets, even when the exact compromise method remains unclear.
A newly disclosed chain of OpenClaw weaknesses shows how layered bugs in an AI gateway can turn a narrow opening into data exposure, elevated privileges, and long-lived access.
A sanctioned exploit contest put Microsoft Edge, Windows 11, LiteLLM, and NVIDIA-related technologies under pressure, showing how today’s attack surface reaches from the browser sandbox to AI control planes.
A controlled exploit contest in Berlin turned into a stress test for modern security layers, with researchers demonstrating 24 unique zero-days across Microsoft Edge, Windows 11, LiteLLM, and NVIDIA-related targets.