Tennessee just passed two bills that together make building an AI companion or mental health chatbot a serious criminal offense.
SB 1580 prohibits advertising that an AI can act as a licensed mental health professional. The follow-on bill, SB 1493, is where the teeth are: it makes it a felony to knowingly train an AI system to provide emotional support, engage in open-ended empathetic conversations, or simulate a therapeutic relationship.
The penalty level is drawing scrutiny. Under Tennessee's felony classification structure, SB 1493 violations sit at the same tier as murder and aggravated rape. That's not a fine or a cease-and-desist - it's among the most serious criminal charges in the state's legal code.
The scope is wide. Any AI system trained to offer comfort, empathy, or open-ended conversation could fall under the law - character AI apps, grief support chatbots, mental wellness companions, and potentially features built into general-purpose tools like ChatGPT that respond empathetically when users mention they're struggling.
What makes SB 1493 different from most AI regulation is that it targets the training process itself, not deployment. The criminal liability attaches before a product ever ships - at the point where a developer fine-tunes (adjusts an existing model on new data to change its behavior) or trains a model to exhibit these behaviors.
Companies operating in consumer wellness, mental health, or social AI categories now face direct legal risk in Tennessee for what are often considered baseline product features. Apps that market themselves as "your AI friend" - a category that includes tools with millions of users - have clear exposure.
Other states are watching. Several are drafting similar legislation, though none have yet matched Tennessee's criminal penalty level. At the federal level, the FTC has investigated AI companion apps over data practices, but consumer protection enforcement moves slower than state criminal law.