Google Secure AI Framework (SAIF) and SAIF 2.0
A conceptual framework for secure AI systems, extended in 2025 for agents
Google's practitioner framework for AI security, whose 2025 update set out the three agent security principles now widely adopted: well-defined human controllers, limited powers, and observable actions.
Contribute
Spot an error, or know a record we should add? Members can propose additions and submit corrections.