Last month Sen Hawley (R,MO) introduced S 2938, the Artificial Intelligence Risk Evaluation Act of 2025. The bill would require DOE to establish an Advanced Artificial Intelligence Evaluation Program, and each year submit to Congress a detailed recommendation for Federal oversight of advanced artificial intelligence systems. No new funding is provided in the bill.
Definitions
Section 3 of the bill provides definitions of 14 key terms used in the legislation. Of particular interest are the terms:
Reading these definitions will allow anyone that has seen the Terminator movies to understand where Hawley is going with this legislation. I am moderately surprised that the crafters did not come up with a bill title that would allow for it to be called the CONNOR Act.
Advanced AI Evaluation Program
Section 5 requires DOE to establish an Advanced Artificial Intelligence Evaluation Program. That program would:
Offer standardized and classified testing and evaluation of advanced AI systems to systematically collect data on the likelihood of adverse AI incidents for a given advanced AI system,
Implement testing protocols that match or exceed anticipated real-world AI jailbreaking techniques, including adversarial testing by red teams with expertise comparable to sophisticated malicious actors,
To the extent feasible, establish and facilitate classified, independent third-party assessments and blind model evaluations to maintain transparency and reliability,
Provide participating entities with a formal report based on testing outcomes that clearly identifies evaluated risks and safety measures,
Develop recommended containment protocols, contingency planning, and mitigation strategies informed by testing data to address identified risks,
Inform the creation of evidence-based standards, regulatory options, guidelines, and governance mechanisms based on data collected from testing and evaluations,
Assist Congress in determining the potential for controlled AI systems to reach artificial superintelligence, exceed human oversight or operational control, or pose existential threats to humanity by providing comprehensive empirical evaluations and risk assessments, and
Develop proposed options for regulatory or governmental oversight, including potential nationalization or other strategic measures, for preventing or managing the development of artificial superintelligence if artificial superintelligence seems likely to arise.
Section 4 of the bill would require each covered advanced artificial intelligence system developer to participate in the program. That participation would include providing to the Secretary, on request, materials and information necessary to carry out the program, including:
The underlying code of the advanced artificial intelligence system,
Data used to train the advanced artificial intelligence system,
Model weights or other adjustable parameters for the advanced artificial intelligence system,
The interface engine or other implementation of the advanced artificial intelligence system, and
Detailed information regarding the training, model architecture, or other aspects of the advanced artificial intelligence system.
The bill would prohibit the deployment of an advanced artificial intelligence system for use in interstate or foreign commerce unless the person deploying the system is a participant in the program. Violators would be fined not less than $1 million per day of the violation.
The program would terminate seven years after the bill is enacted.
Permanent Framework
DOE would be required to submit to Congress a detailed recommendation for Federal oversight of advanced artificial intelligence systems, drawing directly upon insights, empirical data, and lessons learned from the program. The plan would:
Summarize and analyze outcomes from testing, identifying key trends, capabilities, potential risks, and system behaviors such as weaponization potential, self-replication capabilities, scheming behaviors, autonomous decisionmaking, and automated AI development capabilities,
Recommend evidence-based standards, certification procedures, licensing requirements, and regulatory oversight structures specifically informed by testing and evaluation data, ensuring alignment between identified risks and regulatory responses,
Outline proposals for automated and continuous monitoring of AI hardware usage, computational resource inputs, and cloud-computing deployments based on observed relationships between those factors and AI system performance or emergent capabilities,
Propose adaptive governance strategies that account for ongoing improvements in algorithmic efficiency and system capabilities, ensuring that regulatory frameworks remain relevant and effective as AI technology advances,
Suggest revisions with respect to Federal oversight or resourcing, such as a new office within an existing agency, a new agency, or additional funding, that may be necessary to develop and administer a permanent framework for oversight of advanced artificial intelligence systems, and
Provide comprehensive evaluations regarding the potential for tested AI systems to exceed human oversight, approach artificial superintelligence, threaten economic competition (including in labor markets), undermine civil liberties, and pose existential risks to humanity, including clearly articulated options for regulatory or governmental oversight measures to address scenarios of imminent concern identified through testing.
Moving Forward
Neither Hawley, nor his sole cosponsor {Sen Blumenthal (D,CT)}, are members of the Senate Commerce, Science, and Transportation Committee to which the bill was assigned for consideration. This generally means that there is essentially no chance that the Committee will consider the bill. The AI industry, and their supporters in the Senate, are certain to oppose the disclosure requirements in this bill. What is not clear at this point is if that opposition is sufficient to block this legislation, if it were considered.
Commentary
It is interesting that the crafters of this bill selected DOE as the agency to conduct this program. While DOE certainly has an interest in the energy requirements to support these advanced AI programs, it is not necessarily the best agency to oversee this proposed program. An argument could be made that CISA would be better suited to conduct program that is essentially a cyber security program. Unfortunately to Hawley that would have required that the bill be primarily assigned the Senate Homeland Security and Governmental Affairs Committee for consideration. With the anti-regulatory Sen Paul (R,KY) currently the Chair of that committee, there is no way that this bill would ever be considered by that committee. So Hawley (who is a member of that Committee as is Blumenthal) used DOE as the oversight agency, relying on their security functions and advanced technology capabilities to provide the justification.
This bill does not actually allow for DOE to conduct regulatory oversight of covered advanced AI system developers or systems, but the requirement for information provision to DOE with the accompanying penalty for noncompliance makes this a de facto regulatory scheme, that would be certain to morph into an active regulatory effort. This is especially important because there is no language in the bill that would prohibit DOE regulatory efforts. More importantly, there is no language that would limit DOE from sharing the information either on its own initiative or under provisions of the Freedom of Information Act.