ollama | AI Security Wire

img of NVIDIA NemoClaw Flaw Let Malicious Websites Hijack Local AI Agents

Oasis Security disclosed CVE-2026-65105, a DNS rebinding flaw in NVIDIA's NemoClaw stack that let a single malicious webpage seize control of a local Ollama server and plant a hidden, persistent instruction inside the model's chat template.

img of llama.cpp: Unauthenticated RCE in HTTP Server and RPC Backend

Aug 17, 2026 5 min read

Vulnerabilities

llama.cpp: Unauthenticated RCE in HTTP Server and RPC Backend

Two high-severity remote code execution vulnerabilities in llama.cpp expose local AI inference stacks to unauthenticated network attack — one through the HTTP completion server via a missing negative-value check, one through the RPC backend's deserialisation logic.

img of Activation Steering Attacks on Open-Source LLMs: What Defenders Need to Know

Jul 24, 2026 7 min read

Research

Activation Steering Attacks on Open-Source LLMs: What Defenders Need to Know

Activation steering bypasses prompt-level safety controls by manipulating a model's internal representations at inference time. It requires white-box access, which makes local open-source LLM deployments the primary exposure surface.

img of CVE-2026-7482: Bleeding Llama Exposes 300,000 Ollama Deployments

Jul 22, 2026 4 min read

Vulnerabilities

CVE-2026-7482: Bleeding Llama Exposes 300,000 Ollama Deployments

A critical heap out-of-bounds read in Ollama's model loader lets unauthenticated attackers drain server memory in three API calls. Around 300,000 internet-facing instances are estimated at risk.

img of LLMjacking Evolves: Attackers Use Exposed Ollama as AI Hacking Engine

Jul 8, 2026 4 min read

News Brief

LLMjacking Evolves: Attackers Use Exposed Ollama as AI Hacking Engine

Sysdig researchers have documented a threat actor weaponising misconfigured Ollama model servers as the reasoning engine for an autonomous multi-stage penetration testing framework called VAPT, marking a significant shift in how stolen AI compute is being used.

img of Exposed AI Gateways Are Being Weaponized to Attack Third Parties

Jun 30, 2026 5 min read

News Brief

Exposed AI Gateways Are Being Weaponized to Attack Third Parties

Zenity Labs has documented attackers exploiting unpatched LiteLLM vulnerabilities and unauthenticated Ollama endpoints to run autonomous pentesting campaigns, steal compute, and exfiltrate data from victim organisations.

Jun 25, 2026 5 min read

Threat Actors

Threat Actors Are Systematically Scanning Enterprise LLM Deployments

GreyNoise honeypots captured 91,403 attack sessions targeting enterprise LLM endpoints across two distinct campaigns between October 2025 and January 2026. One campaign fingerprinted 73+ model endpoints across all major AI providers. The other exploited SSRF vulnerabilities in Ollama and Twilio integrations.

img of LLMjacking Evolved: Stolen Compute Powers Autonomous Offensive Tools

Jun 22, 2026 4 min read

Threat Actors

LLMjacking Evolved: Stolen Compute Powers Autonomous Offensive Tools

Sysdig caught a threat actor using a misconfigured Ollama instance as the reasoning engine for an automated offensive pentesting framework — a significant escalation from credential theft to weaponised AI infrastructure.

img of CVE-2026-31204: Ollama SSRF and File Read

May 22, 2026 4 min read

Vulnerabilities

CVE-2026-31204: Ollama SSRF and File Read

CVE-2026-31204: an SSRF vulnerability in Ollama allows local-network attackers to read arbitrary files and reach internal services via the model pull endpoint. All versions affected.