Classify content against a stated safety policy
Assess whether supplied text appears to violate a supplied safety policy.
Desired outcome
A policy-violation classification signal for downstream review or handling.
Explore from the outcome
Where this function can lead
These documented connections are starting points for selection, not a ranked list.
Model releases
Identifiable models documented for this function.
- ShieldGemma 1
Google text safety moderation models in 2B, 9B, and 27B sizes, built on Gemma 2.
Link evidence: ShieldGemma model card
Typical inputs
- A user prompt or model response and the policy to apply
Typical outputs
- A classification signal, such as a yes/no answer or score
Sources
Last reviewed 2026-09-30