Researchers manipulated ChatGPT and five other commercial AI tools to create malicious code that could leak sensitive information from online databases, delete critical data or disrupt database ...