Since our last update, we have been deep in the world of AI security and authorization best practices. 

🤖 We launched our ebook "Zero Trust for AI: Securing MCP Servers", a direct response to the overwhelming interest from our popular MCP webinar. It covers MCP server security risks and provides a practical blueprint for securing MCP servers using externalized and fine-grained authorization.

We published new resources to help teams strengthen their authorization strategy. These include demos on making application authorization context-aware with Cerbos outputs, using GitOps with Cerbos, getting started with Cerbos schemas, and filtering database results with Cerbos query plans.

Beyond Cerbos specific content, we shared an overview of the best open source auth tools for enterprises, and covered AuthZEN standards, which we believe represents the future of authorization.

We continued our AI security focus with a deep dive into MCP security from a CISO and architect's perspective, exploration of the productivity paradox of AI coding assistants, and the introduction of cerbos-fastmcp middleware, which brings policy-based, fine-grained access control to your FastMCP deployments.

Product Updates

Cerbos Hub

Cerbos Hub is the centralized control plane for every authorization decision across applications, AI agents, services, and workloads.

With the latest updates, organizations can now manage authorization for every identity in their system with full visibility, consistent policy enforcement, and alignment with a Zero Trust strategy.

1. Fine-grained, tenant specific authorization

Enables SaaS companies to let customers define their own roles and rules without hardcoding customizations, with tenant-isolated policy storage and realtime updates within platform defined guardrails.

2. Dynamic policy management at scale

Automates the full lifecycle of authorization policies with programmatic creation, updates, and deployment to save engineering hours and accelerate releases.

3. Scalable non-human identity permission management

Provides centralized, policy-based authorization for microservices, workloads, and AI agents. Implements least privilege by default and maintains unified audit trails for all access decisions.

4. Secure authorization for MCP servers

Controls which AI agents can access which Model Context Protocol tools using context-aware policies. Policies are evaluated per agent, per tool, and per session, with full audit context for compliance and debugging.

Learn about Cerbos Hub
Get started with Cerbos Hub

Cerbos x FastMCP integrations

FastMCP is a popular Python framework for building production-ready Model Context Protocol servers. However, a default FastMCP implementation exposes all tools to all users, creating a significant security risk. To address this, we've introduced the cerbos-fastmcp middleware, which brings policy-based, fine-grained access control to your FastMCP deployments. This allows you to define authorization rules in human-readable YAML policies, completely decoupled from your application code.

Helpful Content

[Ebook] Zero Trust for AI: Securing MCP Servers

MCP servers are now among the most privileged components in your AI stack. 

They represent a fundamental new risk that traditional security controls weren't designed to address, requiring immediate attention from security, compliance, and risk management teams.

This ebook draws from our work with customers implementing AI systems and covers real incidents like the recent Supabase and Asana vulnerabilities.

Inside you will find:

• How MCP servers fit into your broader risk management and compliance framework
• Why MCP servers break the traditional chain of identity in enterprise systems
• How Role-Based Access Control fails in dynamic AI environments
• The externalized authorization architecture (PEP/PDP) that enables Zero Trust for AI systems
• Authorization policies and deployment patterns to transform your MCP server from a “confused deputy” into a trusted co-pilot

Upcoming Events

Meet our team members and check out their talks to get valuable insights:

• Alex Olivier, CPO and Co-Founder, will be attending and speaking at ISC2 Security Congress in Nashville, TN, October 28-23. Along with professionals from Apple and Google, Alex will be covering the topic of “Automating IAM for Compliance, Security, and Business Agility”. If you stop by to listen - let us know what you think!
• Alex will also be speaking at CyberArk Workload Identity Day Zero in Atlanta, GA, November 10, as part of the "Securing the Future: Workload Identity in the Age of AI Agents" panel.
• Finally, Alex will be speaking at the Gartner IAM Summit in Grapevine, TX, December 8-10. The panel is called "Extend Your Identity Providers with OpenID AuthZEN, Achieve Fine-Grained Authorization, and Enable Zero Trust". Feel free to say hi and ask any questions you may have about authorization and ZTA.