PAI-SF / 2 / PAI-SF-ACT-001

Independent Kinematic Bounding

Objective

Ensure kinematic limits (force, torque, speed, position) are enforced independent of the AI decision path.

Control / requirement

Limits enforced at hardware/firmware level, architecturally independent from primary AI compute; configurable only through authenticated process; defaults to most restrictive limits on integrity failure.

Applicability

Any system with powered actuation where exceeding kinematic limits could cause harm.

Expected evidence

Architecture documentation showing independence; limit configuration change logs; enforcement test records including attempts to exceed limits via compromised AI path. [T2/T3]

Assurance expectation

Test evidence that limits hold even when the AI model commands otherwise; formal verification expected for higher-consequence systems.

Dependencies

PAI-SF-ATT-001 (hardware root of trust for limit configuration integrity).

Exclusions

Not applicable to systems without powered actuation.

Maturity / conformance relevance

Expected at all conformance levels; depth of independence scales with consequence.

Ecosystem relationship

Extends GAISSF™ access control/system integrity to physical actuation; provides actuator-incident categories for UAIF™; defines actuator containment for AI-IRF™.

Domain

Domain 2. Actuator and Kinetic Command Safety

Download full Control Catalogue ↓