Model Failover

A fault-tolerance mechanism in AI systems that automatically switches to a backup model when the primary model is unavailable. In multi-agent workflows, improper handling of failover can introduce security risks such as inconsistent review states.