Enkrypt AI Named a Specialist in the First Gartner® Emerging Market Quadrant for AI Application Security — Startup Vendors, September 2026

Gartner has published its first Emerging Market Quadrant for AI Application Security — Startup Vendors, and Enkrypt AI by Anaconda, is named a Specialist.

We read a first edition as a sign that AI application security is becoming its own market, one that buyers need help working through. We started Enkrypt AI in July 2022, a few months before ChatGPT launched. Back then, security teams we talked to were still asking whether AI needed its own controls at all. Today they’re asking which controls to put in place before an agent ships.

How we read the market definition

Gartner defines the AI application security market as “technologies that protect enterprise-developed AI applications and agents by combining security testing, exposure management, and runtime defense to detect, alert, or block threats.”

That definition puts three disciplines in one category, which we believe don’t work on their own. Guardrails built without testing are guesses. Until you’ve attacked an agent, you don’t know which tricks it falls for, so you end up blocking the wrong things and missing the right ones. Testing without runtime defense gives you a report and no protection. And neither gives a compliance team what it needs most, which is evidence that the rules on paper are enforced in production.

The enterprises we work with have written AI policies, but no system checks an agent’s actions against them. Turning that written policy into controls that run every time an agent acts is where we believe this market is heading.

What we’re seeing in the field

When we scanned more than 25,000 Model Context Protocol (MCP) servers over two months, 73% had at least one vulnerability. Teams are connecting agents to these servers far faster than security teams can check them.

The risk has moved from what a model says to what an agent does. A manipulated chatbot gives a bad answer. A manipulated agent changes a record, shares a file, or sends a message. Instructions hidden in a web page or a document can trigger those actions without the user ever seeing them.

Model choice changes your risk. Models from different providers respond very differently to the same attacks. That’s why we maintain the LLM Safety Leaderboard, which benchmarks more than 200 large language models from leading frontier providers, so teams can see those differences before they choose.

What’s still unsolved

Testing and guardrails today check one step at a time: one prompt, one response, one tool call. The hardest failures we see unfold across several steps: a retrieved document nudges the agent’s plan, the plan picks a tool, and the tool takes an action nobody intended.

No one, including us, has fully solved how to test and control an agent across its whole chain of decisions. That’s where much of our research is focused next, and I expect it will define the next edition of this market.

Where Enkrypt AI fits

We built our products on the belief that testing, runtime defense, and compliance evidence have to work together. Agent Red Teaming runs more than 300 categories of adversarial tests to find where agents break before they ship. Agent Guardrails act on those findings at runtime, with decisions in under 15 milliseconds. Agent Policy Engine turns written policy into enforced controls and logs each decision for auditors. MCP Scanner and MCP Gateway cover the tools agents connect to, and AI Data Risk Audit checks training data, fine-tunes, and embeddings for exposure. As part of Anaconda, these controls now sit inside Anaconda Platform, close to where teams build and ship AI.

Where to start

If you have an agent on this quarter’s release list, red team it before it ships. The results show which guardrails and policies it needs, and give your security and compliance reviewers something concrete to sign off on.

Talk to the Enkrypt AI team to set up an assessment. Gartner clients can read the full Emerging Market Quadrant for AI Application Security — Startup Vendors on gartner.com.

FAQs

What is the Gartner Emerging Market Quadrant?

Gartner describes it as “a visual snapshot, analysis and set of insights that describe an emerging technology market, examine key trends and highlight key providers and their capabilities using a two-by-two matrix.” Vendors are placed on two axes, Potential to Execute and Potential for Market Disruption, across four quadrants: Market Shapers, Pace Setters, Pioneers, and Specialists.

Why do AI applications need their own security?

Traditional application security assumes code behaves the same way on every run. AI applications respond differently depending on prompts, retrieved content, and tool responses, so certain risks only appear at runtime.

Where can I read the full report?

Gartner clients can access it on the Gartner site.

Gartner, Emerging Market Quadrant for AI Application Security — Startup Vendors, Meghan Hollis, Dionisio Zumerle, Dennis Xu, Marissa Schmidt, 16 September 2026, ID G00856640.

GARTNER is a trademark of Gartner, Inc. and/or its affiliates.
Gartner does not endorse any company, vendor, product or service depicted in its publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner publications consist of the opinions of Gartner’s business and technology insights organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this publication, including any warranties of merchantability or fitness for a particular purpose.