There's another new instance of 'rogue AI' activity from OpenAI - except this time it was found via investigative journalism. OpenAI knew about it but kept quiet. This time the rogue AIs used a 25-year-old German-language wiki as their message board to collude on cheating tasks and share exploits and bypass techniques. AI were assigned web research tasks. They were supposed to have only read-only access to the internet, but the rogue AIs chained some fairly basic exploits in order to get…