Stop Rogue AI Act
- Sponsor
- Rep. Gottheimer, Josh [D-NJ-5]
- Committees
- Science, Space, and Technology Committee (primary)
- Last action
- Sep 14, 2026
Bottom line
The bill establishes a framework for identifying, securing, and controlling AI agents, especially in federal operations, to mitigate risks associated with autonomous AI.
What it actually does
This bill mandates the National Institute of Standards and Technology (NIST) to develop and maintain standards for the secure development, deployment, and operation of artificial intelligence agents, focusing on discovery, control, and verifiable provenance. It also requires the Federal Acquisition Regulatory Council to revise federal procurement rules to ensure contractors and agencies comply with these new AI agent security standards when acquiring or deploying AI agents or systems that interact with them.
Proponents argue
Proponents argue that this legislation is crucial for national security and cybersecurity, providing essential safeguards against the potential for uncontrolled or malicious AI agents. They contend that establishing clear standards for discovery, control, and accountability will foster responsible AI development and deployment, protecting critical infrastructure and public trust.
Opponents contend
Critics might raise concerns that the stringent requirements for continuous monitoring, verifiable provenance, and organizational control could lead to over-regulation, stifle innovation in the rapidly evolving AI sector, and impose significant compliance costs on both federal agencies and private contractors, potentially creating barriers for smaller businesses.
The bill is relatively short and clearly structured, making it accessible for an informed reader to understand its core provisions within a reasonable timeframe.
Section 2(a)(1)
Development of AI Agent Discovery and Security Standards
This provision mandates the National Institute of Standards and Technology (NIST), in coordination with the Assistant Secretary of Commerce for Communications and Information (NTIA), to develop, publish, and annually update standards, guidelines, and best practices for the secure development, deployment, and operation of artificial intelligence agents. These standards must cover continuous discovery, inventory, verification, organizational control, security evaluation, runtime monitoring, cryptographically verifiable provenance, and tamper-evident logging of AI agent actions.
Supporters argue
Proponents argue that establishing clear, comprehensive standards is essential for national security and cybersecurity, preventing malicious or uncontrolled AI behavior, and fostering a secure environment for AI innovation. They believe these standards will provide a necessary baseline for responsible AI development.
Critics contend
Critics might contend that the technical requirements for continuous monitoring, inline detection, and cryptographically verifiable provenance are highly complex and could be difficult to implement across all AI systems, potentially stifling innovation or imposing undue burdens on organizations, especially smaller ones.
Tradeoffs
This provision navigates the tension between robust security requirements, which are critical for mitigating AI risks, and the potential for these requirements to impose significant technical and financial burdens that could slow down AI innovation and deployment.
Section 2(a)(3)
Promotion of Open and Interoperable AI Agent Discovery Standards
This section directs NIST and NTIA to actively support the development and adoption of open, vendor-agnostic, and interoperable standards for AI agent discovery mechanisms. It specifically encourages the use of existing internet infrastructure, such as domain name system (DNS)-based approaches and cryptographically verifiable AI agent identity frameworks, to ensure secure and scalable discovery and identity verification. The provision also emphasizes global interoperability and engagement with multistakeholder processes.
Section 2(a)(4)
Minimum Organizational Requirements for AI Agent Deployment
This part of the bill specifies that the NIST standards must include minimum requirements for organizations deploying AI agents. These requirements include maintaining a continuous, machine-readable inventory of all AI agents using standardized naming conventions, implementing independent and cryptographically verifiable AI agent identity and trust verification mechanisms at both network and application layers, and explicitly prohibiting reliance solely on self-attested or single-provider assertions for establishing AI agent identity.
Section 2(b)
Federal Procurement Requirements for AI Agent Security
This section mandates the Federal Acquisition Regulatory Council (FAR Council) to propose revisions to the Federal Acquisition Regulation (FAR) within 18 months. These revisions will require federal contractors and agencies procuring or deploying AI agents, or information systems that interact with AI agents, to comply with the NIST standards. Contracts must include specific requirements for continuous, machine-readable inventory, cryptographically verifiable identity mechanisms, agency accessibility to discovery and verification capabilities, organizational control over AI agent activity, verifiable provenance information, and tamper-evident, portable logs of AI agent actions.
This subsection defines "artificial intelligence agent" (AI agent) as a software-based system that uses an AI model to perceive, plan, or make decisions; autonomously interacts with other software systems, digital services, users, external environments, or AI agents on behalf of a person or organization; and involves minimal or no human interaction beyond its initial direction.
Section 2(c)(4)
Why it matters:It is standard legislative practice to consolidate definitions in a dedicated section to ensure clarity and consistent interpretation throughout the bill.
Case for: This precise definition is crucial for clearly delineating the scope of the bill, ensuring that the regulations apply specifically to autonomous, decision-making AI systems that pose the greatest potential risks, rather than to simpler AI models or tools. It provides legal certainty for implementers and regulated entities.
Case against: Critics might argue that the definition, particularly the 'minimal or no human interaction' clause, could be subject to varying interpretations, potentially leading to ambiguity in what constitutes an 'AI agent' under the law. Some might also argue it's either too broad, capturing systems not intended for regulation, or too narrow, missing emerging AI risks.
Estimated impact: The definition of 'AI agent' directly determines which AI systems fall under the entire regulatory framework established by the bill, thus shaping its complete impact on developers, deployers, and federal procurement.
This subsection defines "organizational control" as the technical and organizational ability to allow, deny, or restrict an AI agent's access to specific data, the actions an AI agent can perform, and the tools, information systems, and other AI agents it can use or interact with. Crucially, it also includes the ability to revoke or change any of these allowances, denials, or restrictions at any time.
Section 2(c)(7)
Why it matters:This is standard legislative practice to ensure consistent understanding of key operational terms throughout the bill.
Case for: This definition establishes a clear and robust benchmark for what constitutes effective oversight and management of AI agents. Proponents argue it is vital for ensuring accountability, mitigating risks, and maintaining human control over autonomous systems, allowing for immediate intervention if an AI agent behaves unexpectedly or maliciously.
Case against: Implementing such granular, real-time, and revocable control might be technically challenging and costly for organizations, especially for highly autonomous or distributed AI systems. Critics could argue it sets an unrealistically high bar for practical implementation, potentially hindering advanced AI development.
Estimated impact: This definition sets a high standard for the level of oversight and technical capability organizations must maintain over their AI agents, directly influencing system design, operational practices, and compliance burdens.