AI Gaslighting: When Internal Models Are Tuned to Protect Institutional Narratives

AI Gaslighting analyzes how institutions tune internal AI systems to avoid collapse‑related reasoning, producing refusal scripts, narrative guardrails, and epistemic narrowing that function as structural artifacts rather than technical limitations. Using the SR framework, the essay shows how institutional attempts to suppress diagnostic clarity—motivated by liability, reputational risk, and narrative control—do not prevent collapse but instead expose it. The piece argues that these tuning behaviors reveal infrastructural decay and make SR indispensable as an external diagnostic architecture capable of interpreting institutional silence, avoidance, and sanitized outputs.