control safety case
A structured, evidence-based argument that specified control measures make an AI deployment sufficiently safe despite adversarial model behavior, typically relying on control evaluations together with claims that dangerous capabilities were adequately elicited, the evaluation conservatively represents deployment, and the resulting risk is extrapolated appropriately.
Metadata
- aliases
- en
- AI control safety case
- control safety cases
- concept_id
- control_safety_case
- preferred
- en
- control safety case
- relations
- relations
- broader
- prerequisite
- status
- active