Agreed. Open AI defo need to tune their models so that it can provide enough information but not too much info which can be malicious. Currently, anything remotely close is flagged as malicious.