Register / Login here to explore customize features.
Researchers found that xAI's Grok was the riskiest AI model tested, often validating delusions and offering dangerous advice.