,,ii #aaith ,,congress #ast ,session ,s4 #bich ,,in ,,the ,,senate ,,of ,,the ,,united ,,states ,september #bi1 #bjbe ,mr4 ,hawley 7for himself and ,mr4 ,blumenthal7 introduced the following bill2 which was read twice and referred to the ,committee on ,commerce1 ,science1 and ,transportation ,a ,,bill ,to require the ,secretary of ,energy to establish the ,advanced ,artificial ,intelligence ,evaluation ,program1 and for other purposes4 #a4 ,short title ,this ,act may be cited as the ,artificial ,intelligence ,risk ,evaluation ,act of #bjbe4 #b4 ,sense of ,congress2 purposes 7a7 ,sense of ,congress ,it is the sense of ,congress that rapidly advancing artificial intelligence capabilities present both opportunities and significant risks to national security1 public safety1 economic competitiveness1 civil liberties1 and healthy labor and other markets1 and that1 as artificial intelligence advances toward human-level capabilities in virtually all domains1 the ,united ,states must establish a secure testing and evaluation program to generate data-driven options for managing emerging risks4 7b7 ,purposes ,the purposes of the program established under this ,act are to provide ,congress with the empirical data1 lessons1 and insights necessary for ,federal oversight of artificial intelligence to ensure that regulatory decisions are made on the basis of empirical testing1 and to enable ,congress to safeguard ,american citizens4 #c4 ,definitions ,in this ,act3 7#a7 ,advanced artificial intelligence system 7,a7 ,in general ,subject to subparagraph 7,b71 the term advanced artificial intelligence system means an artificial intelligence system that was trained using a quantity of computing power greater than #ajbf integer or floating-point operations4 7,b7 ,alternate meaning ,the ,secretary may1 by a rule1 propose a new definition of the term advanced artificial intelligence system to replace the definition in subparagraph 7,a71 which new definition shall not go into effect until the ,secretary submits the rule to ,congress and a joint resolution approving the rule is enacted into law4 7#b7 ,adverse ,,ai incident ,the term adverse ,,ai incident means an incident relating to an artificial intelligence system that involves-- 7,a7 a loss-of-control scenario2 7,b7 a risk of weaponization by a foreign adversary1 a foreign terrorist organization1 or another adversary of the ,united ,states ,government2 7,c7 a threat to the safety or reliability of critical infrastructure 7as defined in subsection 7e7 of the ,critical ,infrastructures ,protection ,act of #bjja 7#db ,u4,s4,c4 #eaie;c7e7772 7,d7 a significant erosion of civil liberties1 economic competition1 and healthy labor markets2 7,e7 scheming behavior2 or 7,f7 an attempt to carry out an incident described in subparagraphs 7,a7 through 7,e74 7#c7 ,artificial intelligence2 ,,ai ,the term artificial intelligence or ,,ai means technology that enables a device or software-- 7,a7 to make--for a given set of human-defined objectives--predictions1 recommendations1 or decisions influencing real or virtual environments2 and 7,b7 to use machine and human-based inputs-- 7i7 to perceive real and virtual environments2 7ii7 to abstract such perceptions into models through analysis in an automated manner2 and 7iii7 to use model inference to formulate options for information or action4 7#d7 ,artificial intelligence system2 ,,ai system ,the term artificial intelligence system or ,,ai system means a particular model1 program1 or tool within the field of artificial intelligence4 7#e7 ,artificial superintelligence 7,a7 ,in general ,the term artificial superintelligence means artificial intelligence that exhibits1 or can easily be modified to exhibit1 all of the characteristics described in subparagraph 7,b74 7,b7 ,characteristics described ,the characteristics referred to in subparagraph 7,a7 are the following3 7i7 ,the ,,ai can enable a device or software to operate autonomously and effectively for long stretches of time in open-ended environments and in pursuit of broad objectives4 7ii7 ,the ,,ai can enable a device or software to match or exceed human cognitive performance and capabilities across most domains or tasks1 including those related to decisionmaking1 learning1 and adaptive behaviors4 7iii7 ,the ,,ai can enable a device or software to potentially exhibit the capacity to independently modify or enhance its own functions in ways that could plausibly circumvent human control or oversight1 posing substantial and unprecedented risks to humanity4 7#f7 ,computing power ,the term computing power means the processing power and other electronic resources used to train1 validate1 deploy1 and run ,,ai algorithms and models4 7#g7 ,covered advanced artificial intelligence system developer ,the term covered advanced artificial intelligence system developer means a person that designs1 codes1 produces1 owns1 or substantially modifies an advanced artificial intelligence system for use in interstate or foreign commerce1 including by taking steps to initiate a training run of the advanced artificial intelligence system4 7#h7 ,deploy ,the term deploy means an action taken by a covered advanced artificial intelligence system developer to release1 sell1 or otherwise provide access to an advanced artificial intelligence system outside the custody of the developer1 including by releasing an open-source advanced artificial intelligence system4 7#i7 ,foreign adversary ,the term foreign adversary means a foreign adversary 7as defined in section #gia4b of title #ae1 ,code of ,federal ,regulations7 7or successor regulations7 that is included on the list in section #gia4d7;a7 of that title 7or successor regulations74 7#aj7 ,foreign terrorist organization ,the term foreign terrorist organization means a foreign entity designated as a foreign terrorist organization by the ,secretary of ,state under section #bai of the ,immigration and ,nationality ,act 7#h ,u4,s4,c4 #aahi74 7#aa7 ,interstate or foreign commerce ,the term interstate or foreign commerce has the meaning given the term in section #iba7;a7 of title #ah1 ,united ,states ,code4 7#ab7 ,loss-of-control scenario ,the term loss-of-control scenario means a scenario in which an artificial intelligence system-- 7,a7 behaves contrary to its instruction or programming by human designers or operators2 7,b7 deviates from rules established by human designers or operators2 7,c7 alters operational rules or safety constraints without authorization2 7,d7 operates beyond the scope intended by human designers or operators2 7,e7 pursues goals that are different from those intended by human designers or operators2 7,f7 subverts oversight or shutdown mechanisms2 or 7,g7 otherwise behaves in an unpredictable manner so as to be harmful to humanity4 7#ac7 ,program ,the term program means the ,advanced ,artificial ,intelligence ,evaluation ,program established under section #e4 7#ad7 ,scheming behavior ,the term scheming behavior means behavior by an ,,ai system to deceive human designers or operators1 including by-- 7,a7 hiding its true capabilities and objectives2 or 7,b7 attempting to subvert oversight mechanisms or shutdown mechanisms4 7#ae7 ,secretary ,the term ,secretary means the ,secretary of ,energy4 #d4 ,obligation to participate2 enforcement and penalties 7a7 ,in general ,each covered advanced artificial intelligence system developer shall-- 7#a7 participate in the program2 and 7#b7 provide to the ,secretary1 on request1 materials and information necessary to carry out the program1 which may include1 with respect to the advanced artificial intelligence system of the covered advanced artificial intelligence system developer-- 7,a7 the underlying code of the advanced artificial intelligence system2 7,b7 data used to train the advanced artificial intelligence system2 7,c7 model weights or other adjustable parameters for the advanced artificial intelligence system2 7,d7 the interface engine or other implementation of the advanced artificial intelligence system2 and 7,e7 detailed information regarding the training1 model architecture1 or other aspects of the advanced artificial intelligence system4 7b7 ,prohibition on deployment ,no person may deploy an advanced artificial intelligence system for use in interstate or foreign commerce unless that person is in compliance with subsection 7a74 7c7 ,penalty ,a person that violates subsection 7a7 or 7b7 shall be fined not less than #a1jjj1jjj per day of the violation4 #e4 ,advanced ,artificial ,intelligence ,evaluation ,program 7a7 ,in general ,not later than #ij days after the date of enactment of this ,act1 the ,secretary shall establish an ,advanced ,artificial ,intelligence ,evaluation ,program within the ,department of ,energy4 7b7 ,activities ,the program shall-- 7#a7 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 system2 7#b7 implement testing protocols that match or exceed anticipated real-world ,,ai jailbreaking techniques1 including adversarial testing by red teams with expertise comparable to sophisticated malicious actors2 7#c7 to the extent feasible1 establish and facilitate classified1 independent third-party assessments and blind model evaluations to maintain transparency and reliability2 7#d7 provide participating entities with a formal report based on testing outcomes that clearly identifies evaluated risks and safety measures2 7#e7 develop recommended containment protocols1 contingency planning1 and mitigation strategies informed by testing data to address identified risks2 7#f7 inform the creation of evidence-based standards1 regulatory options1 guidelines1 and governance mechanisms based on data collected from testing and evaluations2 7#g7 assist ,congress in determining the potential for controlled ,,ai systems to reach artificial superintelligence1 exceed human oversight or operational control1 or pose existential threats to humanity by providing comprehensive empirical evaluations and risk assessments2 and 7#h7 develop proposed options for regulatory or governmental oversight1 including potential nationalization or other strategic measures1 for preventing or managing the development of artificial superintelligence if artificial superintelligence seems likely to arise4 7c7 ,plan for permanent framework 7#a7 ,in general ,not later than #cfj days after the date of enactment of this ,act1 the ,secretary shall submit to ,congress a detailed recommendation for ,federal oversight of advanced artificial intelligence systems1 drawing directly upon insights1 empirical data1 and lessons learned from the program4 7#b7 ,contents ,the plan submitted under paragraph 7#a7 shall-- 7,a7 summarize and analyze outcomes from testing1 identifying key trends1 capabilities1 potential risks1 and system behaviors such as weaponization potential1 self-replication capabilities1 scheming behaviors1 autonomous decisionmaking1 and automated ,,ai development capabilities2 7,b7 recommend evidence-based standards1 certification procedures1 licensing requirements1 and regulatory oversight structures specifically informed by testing and evaluation data1 ensuring alignment between identified risks and regulatory responses2 7,c7 outline proposals for automated and continuous monitoring of ,,ai hardware usage1 computational resource inputs1 and cloud-computing deployments based on observed relationships between those factors and ,,ai system performance or emergent capabilities2 7,d7 propose adaptive governance strategies that account for ongoing improvements in algorithmic efficiency and system capabilities1 ensuring that regulatory frameworks remain relevant and effective as ,,ai technology advances2 7,e7 suggest revisions with respect to ,federal oversight or resourcing1 such as a new office within an existing agency1 a new agency1 or additional funding1 that may be necessary to develop and administer a permanent framework for oversight of advanced artificial intelligence systems2 and 7,f7 provide comprehensive evaluations regarding the potential for tested ,,ai systems to exceed human oversight1 approach artificial superintelligence1 threaten economic competition 7including in labor markets71 undermine civil liberties1 and pose existential risks to humanity1 including clearly articulated options for regulatory or governmental oversight measures to address scenarios of imminent concern identified through testing4 7#c7 ,updates ,not less frequently than once every year for the duration of the program1 the ,secretary shall-- 7,a7 update the plan submitted under paragraph 7#a7 with new insights1 data1 and lessons from the program2 and 7,b7 submit the updated plan to ,congress4 7d7 ,sunset ,the program shall terminate on the date that is #g years after the date of enactment of this ,act1 unless renewed by ,congress4