AI Frontier Institute

AN ACT to amend Tennessee Code Annotated, Title 4; Title 10, Chapter 7; Title 47; Title 58 and Title 68, relative to artificial intelligence.

View primary source →

Tennessee SB 2171 / HB 1898, the "Artificial Intelligence Public Safety and Child Protection Transparency Act," has passed the House (94-0) as amended and been received by the Senate. It has not been enacted. If enacted, it would add a new chapter (68-107-101 to 68-107-107) to Title 68. Large frontier developers (annual revenue of $500 million or more, training models above 10^26 operations) would have to publish a public safety plan on catastrophic risks (68-107-103(a)(1)). These risks are mass-casualty or $1 billion-plus harms, such as CBRN weapon help, autonomous cyberattacks or crimes, or evading control. Large chatbot providers (revenue of $25 million or more, at least 1 million monthly users, foreseeably accessed by minors) would have to publish a child safety plan (68-107-103(a)(2)). Both would have to publish material plan changes with a justification within 30 days (68-107-103(b)). Before deploying a new or substantially modified model, they would have to publish risk-assessment summaries (68-107-103(c), (d)). Materially false or misleading statements about covered risks or plan compliance would be barred, with a good-faith, reasonable-statement exception (68-107-103(e)). Redactions for trade secrets, security or law would be allowed, with unredacted copies kept for 5 years (68-107-103(f)). The attorney general would set up an incident-reporting channel (68-107-104(a)). Frontier developers would report critical safety incidents within 15 days, or within 24 hours to appropriate authorities if there is imminent risk of death or serious injury (68-107-104(b), (c)). Chatbot providers would report child safety incidents within 15 days (68-107-104(d)). Large frontier developers would also give quarterly summaries of internal-use catastrophic-risk assessments starting January 1, 2027 (68-107-104(f)). The Department of Safety would designate federal laws or guidance that companies may declare they follow to satisfy reporting duties (68-107-104(h)–(j)). Civil penalties would be up to $1 million for a first violation and up to $3 million for each later violation by large frontier developers. Large chatbot providers would face up to $50,000 per violation (68-107-105(a), (b)). Only the attorney general could enforce (68-107-105(c)). Incident notifications and assessment summaries would be exempt from public inspection (Section 2, amending 10-7-504). The act is severable (Section 3) and takes effect January 1, 2027 (Section 4).

← Back to the tracker