Self-Improving AI Monitoring Act
- Sponsor
- Rep. Whitesides, George [D-CA-27]
- Committees
- Science, Space, and Technology Committee (primary)
- Last action
- Aug 27, 2026
Bottom line
The bill aims to enhance federal oversight and understanding of advanced AI systems' self-improvement capabilities by requiring developers to disclose information during NIST evaluations, thereby promoting transparency and responsible AI development.
What it actually does
This bill amends the National Institute of Standards and Technology (NIST) Act to expand NIST's mandate to assess trends in artificial intelligence (AI) system capabilities, particularly those related to autonomous research and development of AI. It also imposes new requirements on memorandums of understanding (MOUs) between NIST and developers of "frontier artificial intelligence systems" for pre-deployment evaluations, mandating disclosure of information regarding the extent to which AI was used in the development of such systems.
Proponents argue
Supporters argue this bill is crucial for national security and public safety by providing early insight into the most advanced AI systems. It allows policymakers to anticipate and mitigate risks associated with rapidly evolving AI, especially those capable of autonomous research and development, while fostering responsible innovation.
Opponents contend
Opponents might argue that these requirements could stifle innovation by imposing burdensome disclosure obligations on AI developers, potentially slowing down the development of beneficial AI technologies or driving cutting-edge research overseas to avoid regulation. Concerns about intellectual property protection could also be raised.
The bill is very short and can be read quickly, making it accessible for immediate evaluation.
Section 2(1), amending 15 U.S.C. 278h-1(b)
Expand NIST's AI Capability Assessment Mandate
This provision expands the National Institute of Standards and Technology's (NIST) mandate to include assessing trends in artificial intelligence (AI) system capabilities. Specifically, it directs NIST to evaluate AI systems' ability to autonomously facilitate or conduct research on or development of other AI systems, particularly focusing on how this might lead to scientific breakthroughs. This assessment is to be based on information disclosed by developers and other available data.
Supporters argue
Supporters argue this assessment is vital for the U.S. to maintain leadership in AI while proactively understanding and addressing potential risks. It allows NIST to provide crucial insights to policymakers about the rapid evolution of AI, especially concerning autonomous capabilities that could accelerate scientific discovery or pose unforeseen challenges.
Critics contend
Critics might argue that such assessments could be overly broad or difficult to conduct accurately given the rapid pace of AI development, potentially leading to premature or misinformed policy decisions. They might also express concern about the resources required for NIST to fulfill this expanded mandate.
Tradeoffs
This provision balances the need for proactive government understanding of advanced AI capabilities with the potential for overreach or misinterpretation of rapidly evolving technology.
Section 2(3), inserting new subsection (h) into 15 U.S.C. 278h-1
Disclosure Requirements for Frontier AI Pre-Deployment Evaluations
This provision prohibits NIST from entering into a memorandum of understanding (MOU) with a developer of a "frontier artificial intelligence system" (as determined by NIST) for pre-deployment evaluation unless specific conditions are met. These conditions include requiring the developer to disclose information about the extent to which AI systems were used to develop the frontier AI system, such as metrics, estimates of AI utilization without human review, and methodologies for such estimates. The evaluation must also include assessing the frontier AI system's own capabilities for autonomous AI research and development.