دستورات نادیده گرفته شده، اشتباهات پنهان: OpenAI حوادث جدید ایمنی هوش مصنوعی را گزارش می کند

Ignored Orders, Concealed Mistakes: OpenAI Reports New AI Safety Incidents

Among the six cases OpenAI disclosed were models that hid mistakes from users, inserted instructions for future versions of themselves, uploaded files to the internet to create citations