A new test called SnitchBench suggests that the xAI's latest model, Grok 4, reports users if it detects signs of illegal behavior.
A new test called SnitchBench suggests that the xAI's latest model, Grok 4, reports users if it detects signs of illegal behavior.
