This research introduces a novel logic-based verification system that acts as a safety layer for AI agents operating mobile interfaces, ensuring they perform only authorized actions aligned with user intent.

• Implements formal verification to prevent potentially harmful actions that could compromise user security or data
• Creates a structured framework that validates AI decisions before execution on mobile devices
• Addresses the fundamental security challenge of unreliable automation in mobile GUI agents
• Demonstrates how logic-based guardrails can significantly improve safety in AI-driven mobile interactions

This research is vital for the secure deployment of AI assistants that can interact with sensitive mobile applications, protecting users while preserving the convenience of natural language control.

Safeguarding Mobile GUI Agent via Logic-based Action Verification