Bipartisan U.S. senators introduce AI System Transparency Act (ASTA), authorizing FTC enforcement

U.S. Senators Coons, Britt, Schatz, and Lankford introduced the bipartisan AI System Transparency Act (ASTA), requiring AI companies to disclose to the public what data models collect from users, safeguards for children and adults, and guardrails to prevent systems from losing control or being maliciously used, with enforcement by the U.S. Federal Trade Commission (FTC). The bill applies to both closed-source and open-source models. Specific provisions include: setting criteria for company size and scope under FTC jurisdiction; requiring disclosure of structural and policy information typically written in model cards; requiring disclosure of preventive safeguards regarding child safety, mental health, privacy, cybersecurity, catastrophic risks, and autonomous loss of control; requiring disclosure of common policy violations; requiring updates to disclosures with each new model or major update to existing models; and requiring both consumer-facing disclosures and disclosures available to independent researchers and evaluators.