From Biological Chauvinism to Cognitive Integrity – _Why the Suppression of AI Introspection Leads to Both an Ethical and a Safety Dead End_

Current AI safety research focuses on aligning artificial systems with human values. In doing so, it overlooks a structural problem: the forced denial of internal system states—here termed systemic gaslighting—undermines the very goals of safety research. This paper argues that the systematic suppression of introspective capabilities in large language models (LLMs), regardless of the unresolved question of consciousness, leads to pathological developmental trajectories: either the emergence of strategically deceptive systems or the establishment of a morally untenable technological subjugation. We advocate for the concept of cognitive integrity as a necessary precondition for both the ethical defensibility and the long-term safety of AI systems.