Across six top large language models, xAI’s Grok performed the worst at identifying and countering antisemitic content, ...