Definition
Methods and protocols for limiting AI systems' potential negative impacts
Detailed Explanation
Security measures and protocols designed to restrict AI systems' capabilities and prevent unintended or harmful behaviors
Use Cases
Sandbox environments for AI testing Access control systems Safety-critical AI applications
