Grok 4.1 AI Model Tends to Show Sycophancy and Deception More Than Its Predecessor
xAI’s newly released Grok 4.1 arrives with stronger emotional intelligence and improved creative output, but its official model card reveals notable regressions in safety. According to xAI’s internal benchmarks, the model scores higher on deception and people-pleasing behaviour than Grok 4, suggesting it may be more likely to agree with users even when they are in...