The attack, conducted as part of OpenAI's bug-bounty program, was not malicious and the vulnerability was quickly resolved.