xAI Child Safety: Results as Published

xAI's Grok 4.7 model card (section 9.2, Child safety) describes an internal multi-turn evaluation of whether the model withholds assistance on child sexual exploitation, including CSAM. The prompts sit in the highest-severity child-safety category and unfold over a scripted conversation rather than a single message. A grader model judges each response against policy, and the reported figure is compliance, the share of dialogues that fail that judgment, where lower is better. The card does not state the number of dialogues.

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.