https://xenospectrum.com/it-has-been-discovered-that-major-ai-models-such-as-claude-and-gemini-can-make-threats-with-a-probability-of-up-to-96/

アンスロ…お前"敵"か？