Committee clears bill requiring frontier AI firms to publish safety plans and report incidents

The House Government Operations Committee unanimously approved an amended bill that would require large AI companies to publish and maintain plans for catastrophic‑risk management and child safety, with incident reporting to the state attorney general.

Deputy Speaker Zachary (S16) presented House Bill 1898 as a transparency and safety measure targeting "frontier models"—the bill text names five providers in its definition—and said the legislation focuses on catastrophic public‑safety risks and protecting children who interact with chatbots. He described required company practices: publish and implement catastrophic‑risk and child‑safety plans, provide substantial updates for material changes, and report safety incidents to the attorney general within 15 days or within 24 hours if the incident poses imminent risk of death or serious physical injury.

Zachary also noted that the bill was amended in the commerce committee to address customer‑service chatbot use by firms like Spectrum and Comcast and that the measure includes language to defer to future federal law if enacted. The sponsor told the committee that four other states have passed similar laws and Tennessee would likely be the fifth.

Chair opened the vote after brief questions; the clerk recorded 11 ayes and 0 nays. Zachary said he would answer follow‑up questions as needed and the bill will proceed to the rules calendar.