When does the CA SB 53 take effect?

The California AI Safety Bill (SB 53) took effect on January 1, 2026.

What are the key requirements of the CA SB 53?

Safety testing before initial training and before major updates Publish safety documentation on company website Establish whistleblower protections for employees reporting safety risks

AI Policy Desk

US State·In Effect·Effective January 1, 2026

California AI Safety Bill (SB 53)

Requires developers of large AI models (those trained using more than 10²⁶ FLOPS) to implement safety testing, whistleblower protections for employees who report safety concerns, and incident reporting for safety-critical failures. Narrower in scope than the vetoed SB 1047.

What this means for your team

The FLOP threshold (10²⁶) is extremely high — this law applies to frontier model developers like OpenAI, Anthropic, and Google DeepMind, not to companies building products on top of those models. If you are using APIs from these providers, you are not directly subject to SB 53. However, the law creates indirect accountability: your AI vendors must now have safety testing and incident reporting programs, which improves transparency into the models you rely on.

Key requirements

Safety testing before initial training and before major updates
Publish safety documentation on company website
Establish whistleblower protections for employees reporting safety risks
Incident reporting for safety-critical failures
Applies only to models trained using more than 10²⁶ computing operations

California AI Safety Bill (SB 53)

What this means for your team

Key requirements

Further reading

California AI Safety Bill (SB 53)

What this means for your team

Key requirements

Further reading