AI Cyber Defense Act
- Sponsor
- Rep. Harrigan, Pat [R-NC-10]
- Committees
- Science, Space, and Technology Committee (primary)
- Last action
- Oct 22, 2026
Bottom line
The AI Cyber Defense Act aims to enhance national cybersecurity by developing industry standards and best practices for defending against specific AI attacks and leveraging AI for defensive purposes, while excluding entities from countries of concern.
What it actually does
This bill mandates the National Institute of Standards and Technology (NIST) to engage with private sector artificial intelligence stakeholders to define "artificial intelligence distillation attacks" and develop voluntary standards for their detection and prevention. It also requires NIST to establish an "AI Cybersecurity Roundtable" to create voluntary standards for the defensive use of AI in cybersecurity, with participation from AI stakeholders and critical infrastructure operators.
Proponents argue
Supporters argue that this bill is crucial for national security, as it proactively addresses emerging threats from AI-specific attacks like distillation attacks and promotes the responsible development of AI for defensive cybersecurity measures. By involving private sector experts and critical infrastructure operators, it ensures practical, effective, and widely adopted standards, safeguarding vital systems from foreign adversaries.
Opponents contend
Critics might contend that mandating NIST to define and standardize against specific AI attack types could prematurely lock in definitions or practices that quickly become outdated due to the rapid evolution of AI technology. They may also argue that the exclusion of "entities of concern" could hinder global collaboration on cybersecurity standards, potentially isolating the U.S. from valuable international expertise in a field that requires broad cooperation.
The bill is very concise and can be read and understood quickly, allowing ample time for informed evaluation.
Section 2(a)
NIST Engagement for AI Distillation Attack Standards
This provision directs the National Institute of Standards and Technology (NIST) to work with private sector artificial intelligence companies within 180 days to create a formal definition for "artificial intelligence distillation attacks." Following this, NIST must also develop and publish voluntary standards and best practices for identifying and preventing these specific types of AI attacks. The publication of these standards must occur within one year of their establishment.
Supporters argue
Supporters argue that proactively defining and establishing standards for AI distillation attacks is essential for national security, as these sophisticated attacks pose a significant and evolving threat to data integrity and AI system reliability. This provision ensures that the U.S. remains at the forefront of AI cybersecurity.
Critics contend
Some critics might argue that mandating a definition for a rapidly evolving threat like "AI distillation attacks" could be premature and risk locking in a definition that quickly becomes obsolete, potentially stifling innovation or misdirecting resources.
Tradeoffs
The tension lies between the need for immediate action to address emerging threats and the risk of creating overly rigid standards in a fast-moving technological domain.
Section 2(b)
Establishing the AI Cybersecurity Roundtable for Defensive AI Standards
This provision requires the NIST Director to establish an "AI Cybersecurity Roundtable" within 180 days of the bill's enactment. The Roundtable's primary purpose is to develop voluntary standards and best practices specifically for using artificial intelligence to defend against cybersecurity threats. NIST is directed to ensure representation from private sector AI stakeholders and owners and operators of critical infrastructure on this Roundtable. Within one year of the Roundtable's establishment, NIST must publish these developed standards and best practices on its website.
The definition of "private sector artificial intelligence stakeholders," who are mandated to engage with NIST on AI standards, explicitly excludes any "entity of concern." An "entity of concern" includes foreign persons or entities headquartered in or controlled by countries like China, Russia, Iran, North Korea, and Cuba, with exceptions for U.S. citizens or lawful permanent residents.
Section 2(c)(5)(C)
Why it matters:This is standard legislative practice for defining terms and their limitations. However, its placement in definitions means its significant policy impact on international collaboration might not be immediately obvious.
Case for: Proponents argue this exclusion is vital for national security, preventing adversarial nations or their proxies from influencing U.S. cybersecurity standards or gaining insight into defensive AI strategies. It ensures that the development of critical standards is protected from foreign interference.
Case against: Critics might argue that excluding entities from certain countries, even those deemed "of concern," could limit the pool of expertise available for developing robust global AI cybersecurity standards. Cybersecurity threats are global, and isolating U.S. efforts could lead to less effective or interoperable solutions in the long run.
Estimated impact: This provision significantly shapes the composition of expert groups advising NIST, potentially narrowing the scope of international perspectives in U.S. AI cybersecurity standard development.