Tennessee bill criminalizes conversational AI training

This is the most aggressive anti-AI legislation to date, effectively criminalizing the core alignment techniques used to make modern LLMs helpful and empathetic.

• –Vague definitions of "emotional support" and "mirror interactions" could capture any conversational AI, from ChatGPT to customer support bots.
• –Class A felony classification places AI training in the same category as first-degree murder in Tennessee.
• –The lack of jurisdictional boundaries creates a legal minefield for global developers whose services might be accessed by Tennessee residents via VPN.
• –RLHF alignment, essential for making AI useful and safe, is directly targeted because it imbues models with human-like warmth and conversational nuance.
• –$150,000 per-violation civil penalty creates an immediate and massive liability risk for any AI-integrated SaaS startup.