Privacy & security

Lynx AI’s architecture is structured around local-first execution and bounded authority. Operating at the OS level requires clarity about data handling and system access.

Local by default

Lynx AI runs on your machine. Files and contextual data remain within your operating system environment. There is no hidden virtual desktop or remote mirror of your session.

Explicit permissions

All system access is governed by macOS privacy controls. Lynx AI cannot:

• Bypass permission prompts
• Access unapproved folders
• Escalate privileges

You retain authority over:

• Microphone access
• Calendar integration
• Folder scope

Sandbox-prioritized execution

Where possible, actions run in constrained environments. This reduces risk surface and prevents uncontrolled background behavior.

Execution remains:

• Mode-bound
• Permission-bound
• Scope-bound

No surveillance model

Lynx AI does not monetize user behavior. It does not build advertising profiles or sell usage data. Persistent memory remains local and structured.

Intelligence is applied to assist your work — not to extract value from your activity.

Structural trust

Trust is not established through marketing claims but through architectural constraints.

Lynx AI’s design separates:

• Cognition from authority
• Context from surveillance
• Power from escalation

Operating at the system level requires restraint. Lynx AI is structured around it.