Introduction
Everywhere AI is being adopted, new application endpoints appear. From Vibe Coding tools to large-scale model APIs and pipelines, these endpoints are the front doors into your AI usage — and the first places attackers probe.
The problem? Most organizations have no consolidated way to see them all.
That’s why Wiz AI Security now includes a new Application Endpoints widget, powered by the Wiz Dynamic Scanner. It surfaces live, validated endpoints across the entire AI spectrum: AI Security, AI as a Service, AI Tools, AI Pipelines, AI Frameworks & Toolkits, and AI Models.
What You Can See (and Why It Matters)
The new widget doesn’t just show you a list of open services — it reveals the actual entry points into your AI adoption. These are endpoints that represent the real attack surface:
AI Developer Tools : AI-driven coding workflows that may introduce helper APIs or services outside formal review. (Vibe Coding)
AI Pipelines: Endpoints that move data in and out of training and deployment pipelines.
AI as a Service: Cloud AI platform APIs integrated directly into environments.
AI Frameworks & Toolkits: Developer libraries that expose default endpoints.
AI Models: Direct model-serving endpoints where sensitive data flows.
AI Security: Guardrail or governance services that themselves expose endpoints.
MCP Endpoints: Model Context Protocol interfaces that manage agent/server communication.
Unlike a basic scan that says “service exists,” Wiz validates whether these endpoints are live, exposed, and reachable in runtime — then ties them into the Security Graph so you can see what data, identities, and workloads they touch.
From Visibility to Action
The new widget goes beyond visibility — it helps teams take action right from within AI-SPM:
Explore endpoints directly: Visit each surfaced endpoint through a link in the widget to its IP address to investigate exposure in real time.
Trace to the underlying workload: See which workload hosts the technology, review related issues or misconfigurations, and remediate them — closing the loop from discovery to fix.
End-to-end response: Move from visibility to investigation to remediation, all in one place.
Real-World Examples
To make this concrete, here are two scenarios Wiz surfaces today:
MCP Endpoint Exposure
An MCP endpoint left open to the internet is flagged by Wiz. The Security Graph shows it connects into sensitive data stores, turning what looks like an overlooked configuration into a potential breach path.
Vibe Coding Endpoint in Production
A developer experimenting with Vibe Coding spins up a test API for iteration. Wiz surfaces it as live, shows it ties into a pipeline, and highlights that it’s handling sensitive customer data. What started as a helper service could expose critical assets if left unseen.
Why Wiz, Why Now
Most tools capture only one slice of the AI picture — cloud APIs, SaaS plugins, or isolated model endpoints. Wiz is delivers end-to-end AI endpoint visibility in a single place, dynamically validated and fully contextualized across your entire cloud environment.
This gives security teams a true single source of truth for AI endpoints — from Vibe Coding tools to MCP — and the ability to prioritize exposure based on what actually matters: sensitive data access, risky connections, and real attack paths.
AI endpoints are where innovation meets risk. They are the real-world access points into AI tools, services, pipelines, and models — showing what is actually exposed, not just what exists. Endpoints reveal how sensitive data flows, where shadow AI is being adopted, and how AI connects back into the broader cloud environment.
Wiz continuously validates these endpoints and connects them to identities, data, permissions, and infrastructure in the Security Graph, turning raw exposure into actionable, prioritized risk that teams can act on immediately.
Grounded in AI Security Best Practices
Wiz continuously aligns its AI-SPM capabilities with leading security frameworks such as the OWASP Top 10 for LLMs. These guidelines inform how Wiz prioritizes risks, identifies vulnerable endpoints, and helps teams follow a consistent approach to securing their AI environments.
