IV.
Edge detail
Page 1 of 1evaluates_policy
evaluates_policy ledger
a Benchmark evaluates a ContentPolicy (safety benchmark targeting a specific policy — distinct from generic Benchmark.targets which targets capability subjects)
| from | to | to kind |
|---|---|---|
| benchmark:harmbench | content-policy:default-acceptable-use | ContentPolicy |
| benchmark:harmbench | content-policy:eu-ai-act-aligned | ContentPolicy |