Blog

Technical deep dives on AI agent security from the Ring Zero team.

Threat Intelligence··9 min read

How a Prompt Injection Attack Exfiltrated AWS Credentials — Step by Step

A step-by-step reconstruction of a prompt injection attack that read AWS credentials and staged them for exfiltration — and what the EDR saw vs. what it missed.

Prompt InjectionCredential ExfiltrationAttack ChainAWS
AI Agent Security Concepts··7 min read

Why Your EDR Cannot See AI Agent Attacks

CrowdStrike, SentinelOne, and Microsoft Defender were built for human-initiated threats. AI agents create a fundamentally different attack surface that operates in the gap between application and kernel.

EDRAI Agent SecurityCrowdStrikeSentinelOne
Detection Engineering··8 min read

Kernel Signals That Matter for AI Agent Session Monitoring

Not all kernel events are equal for agent security. This guide covers which eBPF hooks, syscalls, and file access patterns actually matter — and which ones are noise.

eBPFKernel MonitoringDetection EngineeringAI Agents
Comparative Analysis··7 min read

CrowdStrike vs. Ring Zero: What Each One Sees During a Copilot Session

A side-by-side comparison of the telemetry CrowdStrike Falcon and Ring Zero produce during a live AI coding agent session — with a simulated prompt injection attack.

CrowdStrikeComparative AnalysisEDRAI Agent Security
Compliance & Governance··5 min read

What SOC 2 Auditors Will Start Asking About Your AI Agents

SOC 2 audits are catching up to agentic AI deployments. Here's what auditors will ask, what evidence they'll expect, and why application-layer logs are not sufficient.

SOC 2ComplianceAI GovernanceAudit
AI Agent Security Concepts··8 min read

Provenance Graphs: Why Syscall Streams Are Not Enough for Agent Security

Event streams show what happened. Provenance graphs show why. Here's how Ring Zero reconstructs causal chains from kernel events to detect attacks that no individual alert would catch.

Provenance GraphGraph RAGDetection EngineeringAI Security