H.R.8094 - AI Foundation Model Transparency Act of 2026
AFMTA2026 · U.S. Congress
Introduced on March 26, 2026, by a bipartisan group of U.S. lawmakers, H.R.8094 would require developers of large AI foundation models to publicly disclose information about training data, model design, known limitations, risks, and evaluation methods. The bill targets developers of large-scale AI models and imposes transparency obligations without directly regulating how those models may be used or deployed. Its stated objective is to enable public scrutiny of foundation model characteristics without placing operational restrictions on AI development.
Overview
H.R.8094 would establish a federal disclosure regime specifically for foundation models, a category of large-scale AI systems trained on broad data and adaptable to a wide range of downstream tasks. Covered developers would be required to publish standardized documentation covering training data provenance, model architecture, intended use cases, known failure modes, and evaluation methodologies used prior to release. The bill is framed as a transparency measure rather than an operational or safety mandate, meaning it does not prohibit specific model capabilities or require pre-market approval. Enforcement mechanisms and the designated federal agency responsible for oversight have not yet been specified in the introduced text. The bill was introduced with bipartisan support, signaling some cross-party appetite for AI disclosure requirements at the federal level. As of the introduction date, the bill remains in early legislative stages and has not been assigned to committee markup or advanced to a floor vote.
Key Requirements
- •Developers of qualifying large AI foundation models must publicly disclose training data sources and curation methodology
- •Disclosure must cover model architecture and design decisions relevant to capability and limitation profiles
- •Known limitations, failure modes, and foreseeable misuse risks must be documented and made publicly available
- •Evaluation and testing methodologies applied before model release must be disclosed
- •Disclosures must meet a standardized format to be defined by the designated federal authority
- •Specific thresholds defining which models qualify as covered foundation models, penalty structures, and compliance timelines are not yet defined in the introduced text