S.Amdt. 6610

This amendment, known as the Blocking Large-scale Adversarial Distillation Efforts Act of 2026 (or BLADE Act), aims to add a new title to the bill to protect closed-source artificial intelligence models owned by United States persons. Its purpose is to address and stop model extraction attacks and the use of fraudulent account network providers by persons of concern.

The amendment adds the following requirements and authorities:

• Assessments and Reporting: It requires the Secretary of Commerce to conduct an assessment to identify persons of concern conducting model extraction attacks or acting as fraudulent account network providers. The Secretary must report these findings to specific congressional committees and maintain a public AI Model Extraction Attackers List.

• Prevention and Coordination: It mandates the publication of best practices to detect and prevent these attacks and establishes a mechanism for AI model owners to share information with the Department of Commerce.

• Accountability and Penalties: It allows for the following actions against entities identified as attackers or facilitators:

  • Adding them to the Entity List maintained by the Bureau of Industry and Security.
  • Allowing the President to block and prohibit transactions involving their property and interests within the United States.

• Scope and Exceptions: The act specifically targets "persons of concern" (such as those from the People's Republic of China or the Russian Federation) while explicitly excluding legitimate research methods that follow an owner's terms of service. It also includes exceptions for humanitarian assistance and authorized U.S. national security activities.