{"check":null,"uid":"0527636acb80c469","title":"Reliable Artificial Intelligence Research Act of 2025","title_generated":false,"country":"США","organ":"Конгресс США","kind":"law","kind_name":"Законодательство","lang":"en","date":"2026-09-24","summary":"Закон требует от секретаря Департамента внутренней безопасности проведения конкурсов с призами для продвижения исследований интерпретируемости и обеспечения устойчивости к атакам искусственных интеллектов. Он определяет термины, устанавливает структуру и критерии оценки конкурсов, предусматривает административное управление ими и отчётность перед Конгрессом.","snippet":"","topics":["Искусственный интеллект"],"status":"ok","error":"","text_len":8369,"versions":1,"url":"https://www.congress.gov/bill/119th-congress/house-bill/10594","first_seen":"2026-10-07","last_checked":"2026-11-16 05:19","relevance":"hit","score":26,"query":"","source_key":"congress_us","verdict":{"relevance":"hit","score":26,"topics":["Искусственный интеллект"],"need_body":3,"authorities":[],"evidence":[{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":9,"ctx":"reliable artificial intelligence research act of 2025","zone":"название","weight":3},{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":352,"ctx":"nce the science of interpretability and to develop adversarial robustness with respect to artificial intelligence products, and for other purposes.  ______________________________________________________","zone":"текст","weight":1},{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":991,"ctx":"nce the science of interpretability and to develop adversarial robustness with respect to artificial intelligence products, and for other purposes.  be it enacted by the senate and house of representativ","zone":"текст","weight":1},{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":1228,"ctx":"in congress assembled,  section 1. short title.  this act may be cited as the ``reliable artificial intelligence research act of 2026''.  sec. 2. definitions.  in this act: (1) adversarial robustness.--","zone":"текст","weight":1},{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":1406,"ctx":"adversarial robustness.--the term ``adversarial robustness'' means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or","zone":"текст","weight":1},{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":1596,"ctx":"estricted, or harmful outputs, while maintaining integrity, reliability, and privacy. (2) artificial intelligence.--the term ``artificial intelligence'' has the meaning given the term in section 5002 of","zone":"текст","weight":1},{"topic":"Искусственный интеллект","term":"artificial intelligence","weak":false,"pos":1633,"ctx":"maintaining integrity, reliability, and privacy. (2) artificial intelligence.--the term ``artificial intelligence'' has the meaning given the term in section 5002 of the national artificial intelligence","zone":"текст","weight":1}],"dropped":[{"topic":"Персональные данные","term":"privacy","weak":true,"pos":1583,"ctx":"incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy. (2) artificial intelligence.--the term ``artificial intelligence'' has the meaning given","why":"одиночное упоминание (нужно 3)"}]},"last_changed":"2026-10-07","meta":{"congress":"119","billType":"HR","number":"10594","policyArea":"Science, Technology, Communications","subjects":[],"latestAction":"2025-12-03 Read twice and referred to the Committee on Homeland Security and Governmental Affairs.","textVersion":"ih"},"source_url":"https://www.congress.gov/bill/119th-congress/house-bill/10594","text":"FULL TEXT\n[Congressional Bills 119th Congress]\n[From the U.S. Government Publishing Office]\n[H.R. 10594 Introduced in House (IH)]\n\n<DOC>\n\n119th CONGRESS\n2d Session\nH. R. 10594\n\nTo require the Secretary of Homeland Security to carry out prize\ncompetitions to advance the science of interpretability and to develop\nadversarial robustness with respect to artificial intelligence\nproducts, and for other purposes.\n\n_______________________________________________________________________\n\nIN THE HOUSE OF REPRESENTATIVES\n\nSeptember 24, 2026\n\nMr. Pfluger (for himself, Mr. Harder of California, Ms. King-Hinds, Mr.\nDowning, and Mr. Webster of Florida) introduced the following bill;\nwhich was referred to the Committee on Homeland Security\n\n_______________________________________________________________________\n\nA BILL\n\nTo require the Secretary of Homeland Security to carry out prize\ncompetitions to advance the science of interpretability and to develop\nadversarial robustness with respect to artificial intelligence\nproducts, and for other purposes.\n\nBe it enacted by the Senate and House of Representatives of the\nUnited States of America in Congress assembled,\n\nSECTION 1. SHORT TITLE.\n\nThis Act may be cited as the ``Reliable Artificial Intelligence\nResearch Act of 2026''.\n\nSEC. 2. DEFINITIONS.\n\nIn this Act:\n(1) Adversarial robustness.--The term ``adversarial\nrobustness'' means the degree to which an artificial\nintelligence model is able to resist attacks that would induce\nit to produce incorrect, restricted, or harmful outputs, while\nmaintaining integrity, reliability, and privacy.\n(2) Artificial intelligence.--The term ``artificial\nintelligence'' has the meaning given the term in section 5002\nof the National Artificial Intelligence Initiative Act of 2020\n(15 U.S.C. 9401).\n(3) Interpretability.--The term ``interpretability'' means\nthe degree to which humans are able to accurately understand\nhow an artificial intelligence model makes decisions and\nconsiders inputs and how the outputs or behaviors of the model\nrespond to a change in the inputs.\n(4) Red-teaming.--The term ``red-teaming'' means a\nstructured, interactive, and adversarial process to test an\nartificial intelligence system by simulating real-world actions\nto find vulnerabilities or flaws in the system.\n(5) Secretary.--The term ``Secretary'' means the Secretary\nof Homeland Security.\n\nSEC. 3. PRIZE COMPETITION FOR ARTIFICIAL INTELLIGENCE INTERPRETABILITY\nRESEARCH.\n\n(a) Prize Competition Required.--Not later than 270 days after the\ndate of enactment of this Act, the Secretary of Homeland Security shall\ncommence carrying out at least 1 prize competition under section 24 of\nthe Stevenson-Wydler Technology Innovation Act of 1980 (15 U.S.C. 3719)\nto advance the science of interpretability in a manner relevant to\ncommercially available or widely used artificial intelligence products.\n(b) Consultation.--In carrying out the prize competition required\nby subsection (a), the Secretary shall consult with--\n(1) the Secretary of Commerce;\n(2) the Director of the National Institute of Standards and\nTechnology;\n(3) the National Cyber Director;\n(4) the Director of the National Science Foundation; and\n(5) any industry expert from the artificial intelligence\nsector in the United States that the Secretary considers\nrelevant.\n(c) Structure and Evaluation Criteria.--\n(1) In general.--The Secretary shall develop the structure\nand evaluation criteria for a prize competition carried out\nunder subsection (a) in accordance with the primary purpose\ndescribed in that subsection.\n(2) Competition structure.--The Secretary may--\n(A) structure a competition under subsection (a)\ninto 1 or more phases, including submission of\ninterpretability frameworks, submission of\ninterpretable artificial intelligence models, and\nunique basic research; and\n(B) open these phases to the same, or to distinct,\ncontestant pools.\n(3) Evaluation considerations.--In developing the\nevaluation criteria for the frameworks, models, or methods\nsubmitted for a prize competition under subsection (a), the\nSecretary shall consider--\n(A) the degree to which a submission advances\nbroadly applicable principles of artificial\nintelligence interpretability;\n(B) the practical value of a submission in making\nartificial intelligence more understandable and\nreliable in high-risk, high-value use cases; and\n(C) the likelihood that the unique research\nsubmitted will create standards for artificial\nintelligence interpretability in the government or\nindustry.\n(d) Program Administration.--The Secretary may enter into\ncontracts, cooperative agreements, or other agreements with for-profit\nor nonprofit entities or State, territorial, local, or Tribal agencies\nto design and administer any prize competition carried out under\nsubsection (a).\n\nSEC. 4. PRIZE COMPETITION FOR ARTIFICIAL INTELLIGENCE ADVERSARIAL\nROBUSTNESS RESEARCH.\n\n(a) Prize Competition Required.--Not later than 270 days after the\ndate of enactment of this Act, the Secretary shall commence carrying\nout at least 1 prize competition under section 24 of the Stevenson-\nWydler Technology Innovation Act of 1980 (15 U.S.C. 3719) to develop\ncapable artificial intelligence models that are designed to exhibit\nadversarial robustness in circumstances necessary for at least 1 high-\nimpact, high-risk application in government or industry.\n(b) Consultation.--In carrying out a prize competition required by\nsubsection (a), the Secretary shall consult with--\n(1) the Secretary of Commerce;\n(2) the Director of the Institute of Standards and\nTechnology;\n(3) the National Cyber Director;\n(4) the Director of the National Science Foundation;\n(5) any industry expert from the artificial intelligence\nsector in the United States that the Secretary considers\nrelevant; and\n(6) the head of any Federal agency who has authority or\nexpertise in a high-impact, high-risk application of artificial\nintelligence that could be an appropriate subject for a prize\ncompetition under subsection (a).\n(c) Structure and Evaluation Criteria.--\n(1) In general.--The Secretary shall develop the structure\nand evaluation criteria for a prize competition carried out\nunder subsection (a) in accordance with the primary purpose\ndescribed in that subsection.\n(2) Competition structure.--The Secretary may--\n(A) structure a competition under subsection (a)\ninto 1 or more phases, including submission of\nadversarial robustness frameworks, submission of\nartificial intelligence models, and red-teaming; and\n(B) open these phases to the same, or to distinct,\ncontestant pools.\n(3) Evaluation considerations.--In developing the\nevaluation criteria for the frameworks, models, or methods\nsubmitted for a prize competition under subsection (a), the\nSecretary shall consider--\n(A) the degree to which a submission advances\nbroadly applicable principles of artificial\nintelligence robustness; and\n(B) the practical value of the submission in\nreducing the risk of adversarial attacks in high-risk,\nhigh-value use cases of artificial intelligence.\n(d) Program Administration.--The Secretary may enter into\ncontracts, cooperative agreements, or other agreements with for-profit\nor nonprofit entities or State, territorial, local, or Tribal agencies\nto design and administer any prize competition carried out under\nsubsection (a).\n\nSEC. 5. TRACKING AND REPORTING.\n\n(a) In General.--Not later than 180 days after the date on which\nthe first prize competition concludes, the Secretary shall submit to\nthe appropriate congressional committees a report that includes--\n(1) an evaluation of how the results of the competitions\ninform the fields of interpretability and adversarial\nrobustness;\n(2) an assessment of any gaps in these fields identified by\nthe Secretary over the course of the competitions; and\n(3) any suggested action that Congress should take to\nadvance the fields of interpretability, adversarial robustness,\nand any related research.\n(b) Appropriate Congressional Committees Defined.--In this section,\nthe term ``appropriate congressional committees'' means--\n(1) the Committee on Homeland Security and Governmental\nAffairs of the Senate; and\n(2) the Committee on Homeland Security of the House of\nRepresentatives.\n\nSEC. 6. APPROPRIATIONS.\n\nThere is authorized to be appropriated to the Secretary to carry\nout this section $10,000,000 for the period of fiscal years 2027\nthrough 2031.\n<all>","changes":[],"passport":{"data":{"act":{"jurisdiction":"США","title_official":"Reliable Artificial Intelligence Research Act of 2025","title_short":"","level":"законопроект","date_adopted":"2026-09-24 (внесение)","date_in_force":"","date_version":"","phased":"","status":"законопроект: на рассмотрении","sunset":"","regulator":"Департамент внутренней безопасности","related":""},"goal":{"problem":"развитие науки о интерпретируемости и устойчивость искусственного интеллекта к атакам","goal":"продвижение научных исследований интерпретируемости и устойчивости к атакам в области искусственного интеллекта","targets":"отсутствуют","scope":"коммерчески доступные или широко используемые искусственные интеллектуальные продукты","exclusions":"отсутствует"},"subjects_note":{"protected":"нет"},"subjects":[{"role":"секретарь Департамента внутренней безопасности","who":"секретарь Департамента внутренней безопасности","criteria":"должность","count":""}],"norms":[{"address":"ст. 5 ч. а","addressee":"секретарь Департамента внутренней безопасности","essence":"не позднее 180 дней после завершения первого конкурса секретарь представляет отчет Конгрессу о результатах конкурса и выявленных проблемах","type":"обязанность","mechanism":"отчетность","cost_channel":"содержательный","cost_kind":"регулярный","trigger":"завершение первого конкурса","sanction":"отсутствует","refs":"","form":"цифровая","in_force":"","ru_analog":"требует проверки"}]},"made_by":"GigaChat-2","made_at":"2026-10-07 09:12:28","edited_at":null,"edited_by":null}}