AIRiskAware
AI Governance Glossary
Governance Concept

What Is Responsible Scaling Policy (RSP)?

Responsible Scaling Policy (RSP) is a developer's published framework that ties the deployment of more capable AI to predefined safety evaluations and safeguards.

Definition

Responsible Scaling Policy (RSP)a developer's published framework that ties the deployment of more capable AI to predefined safety evaluations and safeguards.

An RSP (some labs call it a preparedness or frontier safety framework) commits a developer to test models for dangerous capabilities at defined thresholds and to hold back or add safeguards if those thresholds are crossed. It is a voluntary, self-imposed governance mechanism rather than a regulatory requirement, and a reference point for what credible frontier-AI risk management looks like.

Source: Frontier AI developer safety frameworks

Plain-language explanation

An RSP (some labs call it a preparedness or frontier safety framework) commits a developer to test models for dangerous capabilities at defined thresholds and to hold back or add safeguards if those thresholds are crossed. It is a voluntary, self-imposed governance mechanism rather than a regulatory requirement, and a reference point for what credible frontier-AI risk management looks like.

Primary source: Frontier AI developer safety frameworks

Related terms

AI Safety Frontier Model Model Evaluation AI Alignment

See where you stand on AI governance

Take the free 7-question maturity assessment and get a personalised action plan.

Free assessment — 3 minutes →