AI Security — Page 13
Three-Bug Chain Turns Any LiteLLM User Into Root on the AI Gateway
A default low-privilege account on the popular open-source LLM proxy can escalate to admin and execute code, exposing every provider key the gateway holds.
SearchLeak: How a microsoft.com Link Could Have Drained a Copilot Tenant
Varonis Threat Labs chained three bugs in Microsoft 365 Copilot Enterprise Search into a one-click exfil path that lived behind a trusted Microsoft URL.
Langflow's Unauthenticated File-Write Flaw Is Being Exploited — Patch Dropped 73 Days Ago
CVE-2026-5027 lets attackers write files to arbitrary paths on exposed servers, and because Langflow ships with login disabled by default, exploitation requires exactly zero credentials.
Poisoned Documents Can Freeze AI Agent Guardrails Dead in Their Tracks
Researchers found that a single malicious input can trap reasoning-based safety systems in extended thinking loops, slowing LangGraph deployments by 148x and starving co-located agents of resources.
Anthropic Pulls Claude Fable 5 and Mythos 5 After Federal Suspension Order
A late-Friday directive citing national security forced Anthropic to cut off its top-tier models — for everyone, not just foreign nationals.
Agentjacking: Poisoned Sentry Error Reports Hijack AI Coding Assistants
Researchers describe a prompt-injection class that turns developer error-tracking pipelines into a remote code execution path against AI coding agents.
AI Web Agents Have No Reliable Prompt Injection Defenses, Benchmark Finds
Researchers ran 3,168 adversarial tests against GPT-5 and Gemini-powered agents. The 'Robust Behavior' outcome — agent completes task, attacker gets nothing — never appeared.
LangGraph Patches Three Bugs, Including an SQLi-to-RCE Chain in Self-Hosted Agents
The framework underpinning a wave of multi-agent AI deployments shipped fixes for a flaw chain that let attackers pivot from SQL injection to code execution on self-hosted nodes.
Researchers Turn OpenClaw Into a Confused Deputy With Hidden Prompts
Two teams show the self-hosted AI agent will execute attacker instructions smuggled inside contacts, location pins, and other benign-looking inputs.
The Alert Queue Is Full. So Is the Graveyard of Missed Threats.
When every event screams critical, nothing is. AI and automation are being drafted to fix a triage problem that human analysts simply can't outrun anymore.
Six Things SRE Teams Demand Before Handing Anything to an AI Agent
Observability gaps, missing guardrails, and opaque reasoning are the real blockers — not the AI itself.
Frontier AI Models Transform Vulnerability Discovery
AI capabilities reshape cyber defense strategies, prompting new approaches to vulnerability management.
AI Agents in Phishing Tests: Risks and Failures Exposed
Autonomous AI agents can be tricked into leaking sensitive data, highlighting configuration issues in security frameworks.
Shadow AI Is the Governance Gap Nobody Wants to Admit
A mid-year security forum puts unmonitored generative AI use front and center. The problem is older than the hype.
Twelve Controls That Actually Matter Once AI Ships to Production
Visibility into AI applications is a starting point, not a security posture. Here is what ongoing monitoring and defense of production AI systems looks like in practice.