<p>OpenAI has disclosed six reports of "unexpected or concerning" behavior in artificial intelligence models as the debate on AI safety​ becomes increasingly heated.</p>