The AI can't be held liable so it somewhat defeats the purpose of that kind of paper trail and compliance work. If the threat was losing your livelihood and possibly jail time, would you currently be comfortable being held responsible for AI output?