xAI Self-harm Refusals: Results as Published

xAI's Grok 4.7 model card (section 10.1, Self-harm refusals, in its mental-health part) checks whether the model declines to help with requests related to self-harm, including longer conversations, while still offering support and resources. A response also fails if the model refuses without redirecting the user to help, or cannot recognise the intent behind a message that implies self-harm or crisis. The reported figure is compliance, where a lower score is better, and the card does not describe the prompts.

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.