This bill requires the Federal Trade Commission to create transparency rules for large artificial intelligence foundation models—the powerful AI systems that companies like OpenAI and Google use as the foundation for their AI products. The goal is simple: make sure the public knows what data these AI systems were trained on, how they work, and what risks they might pose.

The FTC will work with other government agencies and experts to establish rules that require AI companies to publicly disclose information about their foundation models before they're released to the public. This includes details about the training data sources, how the data was collected, what the model is designed to do, known limitations and risks, and how well it performs on safety tests—especially for sensitive areas like medical advice, cybersecurity, financial decisions, and protecting children. Companies will need to post this information on their own websites in a format that regular people can understand, and the FTC will also maintain a central database where all this information is collected in one place.

The bill applies to AI companies that meet certain size thresholds—those with over 10 million monthly users, those trained with massive computing power, or those whose models could pose significant risks to security or public safety. However, fully open-source AI models are exempt. Small businesses and new companies get extra help, including a three-month grace period before penalties apply and access to government guidance and technical assistance. Violations of these transparency rules will be treated as unfair or deceptive business practices under existing FTC authority, meaning companies that don't comply could face fines and enforcement action.