OpenAI reveals six new cases of AI misbehavior

US artificial intelligence giant OpenAI promised to more systemically report instances of its models going off track, while also publishing six new reports on previously undisclosed incidents of AI misbehavior.