Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

AI coding agents produce authorization logic that compiles, passes review, and enforces the wrong policy. Broken access control ranks first in the OWASP Top 10:2025, where 100% of applications tested showed some form of it, across 1,839,701 recorded occurrences, the highest count of any category on the list. One part of that category is also the part that pattern-based scanning was never built to reach.

Agentic AppSec (agentic application security) is the practice of using a team of AI security agents to run an organization's entire application security program: understanding the application, modeling its threats, finding the vulnerabilities that matter, deciding what is worth fixing, generating and validating fixes, and proving those fixes hold. It applies continuously to both new code and the existing backlog.

Today, we're announcing that Govern Agent Behavior, the capability within Evo Agentic Development Security (ADS) that controls what AI coding agents are allowed to do at runtime, is generally available, starting with MCP Governance.

Every security program has one: a queue of a few thousand findings, or a few hundred thousand, that nobody has worked through and nobody expects to. Most teams file it under technical debt, a cost carried on purpose, paid down when there is room, and tolerable because the interest rate stays low. That accounting held for a long time, because it rested on a single assumption: almost nothing in the queue would ever be reached, or exploited, by anyone.

Evo already knows which AI Assets your teams pulled into your repos, which MCP servers and skills are sitting on your developer machines, which of them carry risk, and which policies they break. Getting to any of it created friction: you leave the tool you are working in, filter a UI, export a CSV, and rebuild the chart you built last quarter, every time it’s needed.

For the past year, I've described this as an AI fog: competing claims, uncertain risks, and leaders struggling to see what mattered. The winds are up now. AI is accelerating software creation, exposing years of accumulated vulnerabilities, and handing both attackers and defenders capabilities that operate at machine speed.

Stopping new security issues in agent-generated code from being deployed is, architecturally, a solved problem. Prevention is the act of keeping a new vulnerability in code from reaching production at any point in the development and release process, including but not limited to preventing its introduction in a feature branch.

Imagine getting three separate security reports back for your new enterprise AI assistant: On paper, the application looks ready for production, but in reality, a threat actor bypasses your guardrails in minutes. How? By using the AI model as an intermediary. The attacker steers the LLM to invoke the internal utility tool, thereby bridging an untrusted prompt directly to the backend execution sink.

Six new security issues for every one issue remediated. That's the ratio Snyk research has found, and it's why the AI Security Engineers Community gave an hour of livestream time to fixing rather than finding. Play Video: Remediation Agents Demystified: Your AI Teammate for Fixing Security Bugs Remediation Agents Demystified paired a fireside chat with a live demo.

We benchmarked how well leading models produce vulnerability fixes that are both secure and functional, across ~150 real vulnerable code samples in JavaScript, Java, and Python. We ran each model on its own and with Snyk Intelligence (the new agentic Agent Fix architecture). The headline findings.