OpenAI’s large language model ChatGPT fixed a security flaw found earlier this week by a researcher within the “Actions” feature of Custom GPTs. Attackers could have exploited a Server-Side Request ...