Facio Blog
Payload-powered product notes, security writing, HITL patterns, and operational guidance from the Facio runtime: long sessions, Placet approvals, audit trails, memory, providers, channels, tools, and Docker-first operations.
Showing 6-10 of 18 articles in Security.
Security
Non-human identities outnumber employees 144:1 in enterprise environments. 78% of employees use unauthorized AI tools. 97% of machine identities have excessive privileges. Shadow AI agents aren't a future threat — they're inside your organization right now, operating without anyone's approval.
Security
Memory poisoning turns prompt injection into a stateful attack. MINJA achieves 95% injection success, AgentPoison exceeds 80% at under 0.1% poison rate, and Gemini's memory feature was compromised through delayed tool invocation. Here's what defense in depth actually requires.
Security
GitGuardian found 29 million new hardcoded secrets in 2025 — 81% more AI service leaks year over year, and 24,000+ secrets in MCP config files alone. When AI agents inherit standing credentials from humans, the blast radius is catastrophic. Here's how to fix it.
Security
Microsoft's security research found that a single prompt can trigger remote code execution in popular AI agent frameworks. Two real CVEs, one attack chain, and what enterprise teams must do before their agents become shells.
Security
A kill switch needs someone watching. At 3 AM, nobody is watching. Circuit breakers — automated controls that detect anomalies and self-terminate — are the missing security layer for autonomous agents, and most teams discover this the hard way.