Akto Recognized as a Leader in SafenAI's Agent Security 50

A year ago, there was no clear structure to AI agent security.

Today, it has one of the clearest pictures yet, the Agentic Security 50, a category map of 50 top AI agent security solutions authored by Brett Liu and published by SafenAI and 60 Degrees Capital. Akto is named in it, and positioned at the center of the Venn diagram - across all three of the pillars the map identifies.

What is the Agentic Security 50 Map?

The Agentic Security 50 is a community-led mapping of the AI agent security market released by SafenAI and Sixty Degree Capital. It catalogs companies building in the category and organizes them across four core pillars:

• Discovery & AISPM - finding the agents, models, and AI assets running across an organization, and managing posture across them

• Governance & Audit - controlling what agents can access, who can use them, and producing the audit trails enterprises need

• Runtime Security - protecting agents while they're running, including against prompt injection, data exfiltration, and unauthorized tool use

• Identity & Access - managing the identities of agents themselves, the credentials they hold, and the access they accumulate as they call tools and other agents

Where Akto sits, and why

Akto is at the center of the Venn - sitting at the intersection of Discovery & SPM, Governance & Audit, and Runtime Security. Out of the 75 solutions considered for this project, only 9 sit across all three.

This isn't a positioning claim Akto made. This placement reflects how Akto is setting the standard for AI Agent security, building comprehensively across discovery, governance, and runtime protection from the ground up.

For AI agent security specifically, that means:

Discovery & SPM - Akto inventories the agents, LLM endpoints, MCP servers, and AI assets running across an organization, surfacing shadow deployments, tracking posture drifts, and giving security teams a live view of their AI attack surface.

Governance & Audit - Akto governs what agents can access, what tools they can call, and what data they can touch, producing the audit trails enterprises need for compliance and incident review, especially as AI systems start making decisions autonomously.

Runtime Security - Akto enforces guardrails in real time - blocking prompt injection, catching data exfiltration in flight, and detecting anomalous agent behavior before it becomes an exposure.

Discovery, governance, and runtime aren't just adjacent problems for AI agents — they share the same substrate. The same agents, the same MCP servers, the same tool calls. Watching them through three different products means losing the shared context that makes any of these layers actually useful.

Akto is grateful to be recognized in the map and grateful to Brett Liu, SafenAI, and 60 Degrees Capital for the work of building it. The category is forming. There's a long way to go.

Explore the map: The Agentic Security 50 →