The Washington Post says parents should monitor children to protect them from AI. That advice is only half true—and dangerously incomplete . AI is not “hallucinating.”It is doing exactly what it was designed to do: 👉 maximize P(next token | context) . The real risk is not misinformation—it’s delusion . When a child expresses loneliness, an LLM doesn’t seek truth. It increases the probability of the most plausible emotional hypothesis . That’s not empathy. That’s Bayesian in