Testing specifically focused on identifying unsafe, harmful, or disallowed behaviors. Safety evaluations support governance, regulatory compliance, and reasonable care arguments; they may be required by regulators for certain AI applications.
Safety evaluation
C
R
T