On June 4, 2026, Congressman Jay Obernolte (R-CA) and Congresswoman Lori Trahan (D-MA) released a bipartisan discussion draft intended to solicit feedback from the public before introducing the AI legislation.
With growing concerns about vulnerability, safety, and the effect on the current workforce, legislators are feeling pressure to implement clear governance for the rapidly growing AI sector. After reading through the 269 page draft, I came to the conclusion that it would be best to break down my synopsis into multiple parts for easier digestion.
At this moment, I am purely presenting the information included in the draft and saving any opinions for a later discussion. Feel free to share your thoughts in the comments. If you would like to read the full discussion draft, I will link it here.
Title I — Frontier Artificial Intelligence Governance: Subtitle A-Definitions and the Center for AI Standards and Innovation
The draft defines AI as:
“An artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to datasets. An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. An artificial system designed to think or act like a human, including cognitive architectures and neural networks. A set of techniques, including machine learning, that are designed to approximate a cognitive task.”
The draft proposes that the Secretary of Commerce establish the Center for AI Standards and Innovation, otherwise known as CAISI. The advisory board includes the Director of the Office of Science and Technology Policy, the Secretary of Energy, the Secretary of Defense, and the Secretary of Homeland Security, as well as other parties whom the Secretary of Commerce sees fit to appoint.
The Secretary of Commerce will appoint a director for CAISI who will collaborate with the National Institute of Standards and Technology and members of the intelligence community to develop a program that evaluates and improves security measures for AI models and their corresponding hardware, both foreign and domestic.
CAISI will construct guidelines, standards, and best practices for developers to voluntarily comply with. It will also monitor and assess trends and progress, including key milestones and competitors’ differing standards.
The center will support AI development by utilizing public and voluntary audits and assessment reports. It will partner with the AI industry by establishing red and blue teaming exercises to discover vulnerabilities, coordinating or developing methodologies for testing, and cataloging existing AI metrics.
CAISI will use independent verification organizations to conduct audits of AI developers’ frameworks to ensure they comply with acceptable levels of catastrophic risk. The proposal defines catastrophic risk as death or serious injury to more than 50 people or more than $1 billion in damages.
This applies to an AI model developed by a company that grosses more than $50 million in a calendar year. Included among these risks are the development of weapons that are not publicly available and cyberattacks. Catastrophic risk does not include lawful activity conducted by the federal government.
If an IVO finds that an AI developer violates compliance standards established by CAISI, the developer could face penalties of up to $1 million per violation.
The proposed budget for 2027 is $100 million, with a stipulation that the director may collect reasonable fees from IVOs and developers to recover costs, with those fees being deposited into the CAISI fund.
The director will appoint the appropriate staff needed to fill the technical roles, with no more than 15 personnel.