How to Comply with Safety Protocols — CA Frontier AI Safety
SB 53 · CA · Effective Jan 1, 2026
In effectImplement safety protocols for frontier AI systems.
Implementation steps
- Define safety policies for development and deployment.
- Implement technical safeguards (e.g., testing, monitoring).
- Document protocols and assign ownership.
- Review and update as models or use cases change.
Evidence needed
- Safety policy
- Technical safeguard documentation
Passing criteria
- Documented protocols in place
- Aligned with regulatory expectations
Penalties: Up to $1M per violation. Enforced by Attorney General.
Get your full compliance roadmap
We map this requirement and every other to your AI systems. Sign up free.
Free compliance checkerorSign up free