AI Security Hub
05AI Security Pillar

Tool and Agent Security

Limit the systems, credentials, permissions, and actions available to AI agents.

Business focus

Control AI actions

Why it matters

Security starts with a defined business boundary

Tool and agent security applies least privilege, approval gates, scoped credentials, and action logging when AI can interact with business systems.

Business risk

A wrong answer becomes a wrong action when an AI agent can send email, change a record, share a file, create an account, or trigger another system.

What this pillar covers
  • Approved-tool allowlists
  • Permission and authorization checks
  • Human approval for higher-risk actions
  • Action logging and accountability
Launch reading list

Start with these practical guides

Each pillar begins with one anchor guide and two supporting articles. Published guides become active automatically as they enter the blog.

Publishing soon

Securing AI Agents: A Practical Business Control Framework

Govern an agent from identity and data access through action and response.

Part of the launch series
Publishing soon

Least Privilege for AI Agents: Give Access by Task, Not Convenience

Scope access to the task, user, data, tool, environment, and time window.

Part of the launch series
Publishing soon

OpenAI API Key Security: Project Controls for Business

Separate, store, monitor, rotate, and respond to exposed application credentials.

Part of the launch series

Questions for leadership

  • Which actions can the agent perform without approval?
  • Does each tool use a separate, scoped service identity?
  • Can the business reconstruct what the agent attempted and changed?
  • Who can pause the agent when behaviour looks wrong?
Practical control checklist

Put the pillar into practice

  1. 1Allow only the tools required for the defined task.
  2. 2Give each agent and environment separate credentials.
  3. 3Apply user and business authorization before every action.
  4. 4Require approval for external or state-changing actions.
  5. 5Log tool calls, parameters, approvals, results, and errors.
Review a real workflow

Turn this pillar into operating controls

Map the data, access, approvals, monitoring, and evidence around one important AI use case before expanding it.

Explore the diagnostic