If I can’t find an answer, I’ll just make one up.

https://www.nature.com/articles/d41586-024-03137-3

Bigger AI chatbots more inclined to spew nonsense — and people don't always realize

Artificial-intelligence models are improving overall but are more likely to answer every question, leading to wrong answers.

Did they ask an AI about the guidelines?

https://www.theregister.com/2024/09/25/doj_ai_compliance_guidance_update/

If your AI does the crime, you'll do the time, warns DoJ

If juggling the extreme cost and hazy ROI of AI weren't enough of a headache, the United States Department of Justice (DoJ) now expects enterprise compliance officers to start weighing the tech's potential for harm – or risk stiff fines if it breaks the law.

Nicole Argentieri, the principal deputy assistant attorney general for the DoJ's criminal division, discussed the changes made to the Evaluation of Corporate Compliance Program (ECCP) guidelines [PDF] in an address to the Society of Corporate Compliance and Ethics earlier this week.

The guidelines detail how DoJ prosecutors should approach criminal investigations and evaluate service providers' effectiveness at preventing criminal behavior. As such, the ECCP effectively functions as a guide for compliance officers looking to avoid the DoJ's ire.