“Word choice 'removed' and 'guardrails' emphasizes loss of safety constraints, triggering concern about AI risks without balanced context.”