3

U.S. Artificial Intelligence Strategy： Internal and External Mechanisms， Challenges and Evolutionary Trends

As and technological revolution and industrial transformation， is reshaping theglobal power dynamics.Major countries are accelerating their strategicdeploymentaround technological innovation，standard-setting，and rule building， making technology increasingly intertwined with geopolitics. Since the U.S.released Preparing for the Future of Artificial Intelligence in 2016，ithasintegrated AI into its national strategic system and continuouslyupgraded it，aiming tomaintain its dominance in the global competition.Meanwhile， theinternalandexternal tensions of the U.S.AI strategy have become increasingly prominent， profoundly influencing its global strategy and future trajectory.

INTERNAL MECHANISMSOFTHEU.S. AISTRATEGY

Through policy and mechanism design， a polycentric governance system，innovationecosystemconstruction， and cutting-edge technology deployment， the U.S. AI strategy not onlymaintainsits technological advantages，butalso servesto provide institutional guarantees and strategic foundations for shaping global AI governance rules.

I. Strategic Layout Driven by MultipleObjectives

The U.S. AI strategy presents a systemofmultiple，mutually-nestedobjectivesencompassing“technological supremacy，securityprioritization，economiccompetitionandvalueexportation.\"Its policy design and strategic planningareadvancedonthisbasis.

First，thecoremotivationof theU.S AI strategyis to sustain its global leadership in technology and to contain China'srise.The 2O21Final Reportof the U.S. National Security Commission on Artificial Intelligence states that China is pursuing long-term strategic deployment through scientificresearch，talentcultivationand investment，requiring theU.S.to“spare no effort”to maintain technological innovationadvantagesanddominate AIdevelopment.Thisreflects theU.S. strategic logic of \"leadership equals security.\"

Second， thedeep integrationofAI into the U.S. military system becomes akeyvariablein future combatsystemtransformation. The U.S.DepartmentofDefense（DoD）established theChiefDigitalandArtificial IntelligenceOffice（CDAO） to deepen AI applications in image recognitionand battlefield decision-making. The 2023 DoDData，Analytics，andAI Adoption Strategy introduced the concept of the \"AI-enabled kill chain，”emphasizing theintegrationofalgorithmsintothe entire combat process to enhance op erational efficiencyand friend-or-foe identification capability.

Third，as a strategic technology， AI has become a key pillar for the U.S. in maintaining its global economic influence.The2022U.S.CHIPSand Science Actdesignated AI asa“national strategic industry，\"aiming to secure dominanceincoreareassuch asalgorithminnovation，intelligent hardware，and data governance. Concurrently，through theU.S.-EUTrade and Technology Council （TTC）， the U.S.is strengthening coordination with the EU on seting AI standards and governing cross-border data flows. This effort seeks to establish a global digital economicrulesframework centered on theU.S.norms.

Fourth， the U.S. AI strategy increasingly reflects global governance objectivesorientedtowardsvalue embeddingandinstitutionalexport. TheUnitedNationsadopted itsfirst global resolution on AI proposed by the U.S.， titled Seizing the OpportunitiesofSafe，Secure，andTrustworthyArtificial Intelligence Systems forSustainable Development. This resolution establishes a governance framework centered on“safe， secure， andtrustworthy\"principles.Through multilateral platformslike theGlobal Partnership on Artificial Intelligence universities（GPAI）， theU.S.isadvancing the construction ofmultilateralmechanisms for“TrustworthyAI.”Atthedomestic level，theWhiteHouse OfficeofSci ence and Technology Policy （OSTP） issuedtheBlueprintforanAIBillof Rights，outlining five core ethical principles and driving the shaping of the global AI governance order through a dual-track approach.

II.FederalCoordinationand Polycentric Governance Structure

OnMay20，2025，inWashington，D.C.，U.S.PresidentTrumpannounceda developmentplanfortheGoldenDomeprojectattheWhiteHouse，acomprehensive andmulti-layereddefensesystemwithspace-basedcomponents.TheU.S.militaryaims to enhance thissystemusingAI.

The U.S. AI strategy employs a governance model characterized by federal leadership，polycentric collaborationandinteragencycoordination. Thisapproachutilizesamulti-tiered institutionalnetworktoensureboththe overall coherence ofAI policiesand specialized division of responsibilities.

Interms of overall coordination， the OSTP serves as the core federal agency， responsibleforstrategyformulation， interagency coordination，and ethical regulation.Its subordinate body， the National Artificial Intelligence Initiative Offce， is responsible for policy collabo rationandimplementation.

In terms of technology Ramp;D and standard-setting，theNationalScience Foundation（NSF） and National Institute of Standards and Technology （NIST） play leading roles. The NSF fosters a research network spanning agriculture， education， and manufacturing through establishing AI research institutes. The NIST develops reliability and safetystandardsforAI systems.

Intermsof securityanddefense， theDoD advances AI deploymentvia theCDAO，implementing initiatives liketheAIRapidCapabilitiesCelland Global Information Dominance Experiment to optimize intelligence processingandcross-service jointoperations.

In terms of data governance and ethical oversight， the National Science and Technology Council and its Select Committee on AI guide federal agencies to setunifiedstandardsforAI Ramp;Dpriorities， data openness， security，and privacy， while supervising ethical compliance.

III. Innovation Ecosystem of Multi-Stakeholder Collaboration

TheU.S.emphasizescollaboration among government， enterprises， and"to drive policy guidance， capital investment， technological Ramp;D， and talent cultivation， creating a dynam icand efficient innovationecosystem.

First，policyand funding.The federalgovernmentauthorizedUSD280 billion through the CHIPS and Science Act to support semiconductors，AI infrastructure，and STEM education， designating AI as a priority alongside quantum computing and advanced manufacturing.In May2023，the OSTPreleasedtheupdatedNational Artificial Intelligence Research and Development Strategic Plan， further outlining nine strategic priorities， such as basic research and interdisciplinary integration， education and training， and infrastructure improvement， to steerfundamentalresearchtowardindustrial applications.

Second，researchinnovation mechanisms. Government， industry， and universities collaborate closely in thisregard.TheDefenseAdvanced Research Projects Agency （DARPA） launched the Artificial Intelligence Exploration program， utilizing a smallgrant， rapid-funding mechanism to supportshort-cycle，high-risk，igh-rewardresearchinareaslikeexplainable AI， energy efficiency optimization，and adversarial machine learning. Google's DeepMindestablishedjointresearch projectswithuniversities likeUC Berkeleytoadvancethetranslationof fundamentaltheoreticalinnovations.

Third， talent development.The U.S.prioritizes both international re cruitmentanddomesticcultivation. It attracts global talent by optimizing policies like the H-1B visa and the STEM Optional Practical Training extension.DisciplinessuchasAI， data science，androboticsare formallyincluded in these programs， providing pathwaysforinternational students to work in theU.S. On top of this， the federal government fundsuniversitiesto establish interdisciplinary AI programs，suchastheNSFNationalAIResearch Institutes'workforce development initiatives，to continuously strengthen the talent pipeline.

Fourth， frontier technology layout. TheU.S.concentratesonareaslike generativeAI andembodied intelligence.IngenerativeAI， itaccelerates military applications for battlefield management，intelligencefusion， and decision-making support，aimingtogainanedgein theAIarms race.InFebruary2025，theSpecial Competitive Studies Project proposed aNational Strategy for Robotics to accelerate AI + robotics applications in industry，military，and logistics，while advancing technological decoupling from China to build a strategic“AI + Robotics\"ecosystem.

EXTERNALDIFFUSIONOFTHEU.S.AI STRATEGY

The U.S.AI strategy focuses not onlyon consolidating technological superioritybutalsoemploysaseries ofexternaldiffusion mechanisms aimed at establishing and reinforcing itshegemonicposition in the global AIdomain.Thesepathwaysmanifest primarilyininstitutionalnormalization and strategicsecuritization.

I. Institutional Normalization Path way：The“SoftLock-in\"Logic of Rule Export

TheU.S.leverages institutional spillover to drive rule-setting， achieving discursive dominance in global AI governance and effecting a “soft lock-in” of itstechnological hegemony. Thismechanism involves leading multilateral norm-setting， building a“trustworthy AI alliance\"，and embedding its technical standards into global market rules.

First， in terms of shaping multilateral mechanisms， the U.S. consistently leadsor deeplyparticipates in international platforms like the G7， OECD，and the GPAI， using institutional channels to promote its “responsible AI\" develop ment path and embed core principles of the U.S. like“human rights protection， democratic values， transparency，and accountability.For example，theOECD AI Principleshave become a widely recognized global governance framework，which theU.S. actively pushes its allies to integrate into their policies and industrialstandards.

Second，in terms ofdiffusing rules and standards， the U.S.leverages the narrative framework of“trustworthyandresponsibleAI\"，anddrives standarddiffusionthroughissueslike \"algorithmic transparency\"，“ethical accountability”，and “fairness auditing\". TheNISTAI RiskManagement Frameworkand itsgenerative AI supplement outline a four-stage approach of \"governance-mapping-measurementmanagement\"，providing compliance guidelines for international cooperation.Theseare disseminated via platformslike theTTC and GPAL

Third， in terms of setting technical standards， the U.S.actively engages in internationalstandardsbodieslikethe InstituteofElectrical and Electronics Engineers （IEEE）and the International Organization for Standardization， seeking to dominate rule-setting in areas such as AI ethics， model transparency， anddata interoperability.For instance， U.S.experts held pivotal roles in drafting the IEEE white paper Ethically Aligned Design，facilitating the insertion of U.S.-style institutional concepts. Byembeddingthesestandardsand technical specifications into international market access systems， the U.S. reinforcesitsinstitutional structural powerand governance influence.

II.StrategicSecuritizationMechanism：The“HardContainment\"Dimensionin TechnologyDiffusion

Beyond promoting“rules spillover\" theexternal expansionof theU.S.AI strategyalso exhibitsapronounced trendofstrategicsecuritization.This mechanismmanifestsintheredefinitionof technology's security boundaries，the intensification of technology containment against specific countries， and the construction of geopolitical technology alliances.

First，AI has been systematically integrated into the U.S. national security strategyframeworkandincludedin the list of \"critical and emerging technologies.”The Export Control Reform Actand the Foreign Investment Risk ReviewModernizationActhavebeen leveraged to strengthen scrutiny over exports and mergers or acquisitions involving AI systems， natural language processing，and machine learning：Concurrently，theCleanNetworkinitiative has expanded its scope beyond telecom and 5G to encompassAI， cloud computing，datacenters，and theInternet of Things. This aims to decouple deep infrastructureinterdependencefrom China and others， constructing a global technological isolationsystembased onideological demarcation.

Second， U.S. technology containmentagainst specific countrieshas become increasingly systematic. In October2O22， the Bureau of Industryand Security （BIS） under theU.S. DepartmentofCommerce imposed export controls prohibiting the sale of high performanceGPUs，advanced computing chips， electronic design automation software，andsemiconductormanufacturing equipment to China. In early 2025，theBISintroducednewregulationsrestricting critical items like AI model weights and continuously updated the EntityList for export control， adding 140 Chinese entitiesin December2024andanother83inMarch2025， primarily targetingadvancedchipsand semiconductor equipment.

Third， at the geopolitical level， the U.S. is committed to building an ideological alliancesystemcentered on AI technology. It leverages platforms like theIndo-PacificEconomicFramework （IPEF），G7Digital and TechMinisters' Meeting， and the QUAD technology cooperationmechanismtoimplement coordinated containment. For instance， the2023G7Digital andTechMinisters'Meeting designated AI as a priorityareafordemocraticcooperation， focusing on raising technical access barriers， controlling norm exports， and managingsupplychains.IPEF'sdigital economypillarexplicitlyaddressesAIdata governance， cross-border data flows， and standard-setting. The U.S.- Japan-Netherlandsagreementrestricting EUV lithography machine exports exemplifiesextendingthistechnosecuritization strategy to the end of the supplychain，aiming tocontain China's breakthroughsin high-end chips.

INTERNAL ANDEXTERNAL TENSIONS INTHEU.S.AISTRATEGY

TheU.S. faces multiple challenges inadvancing itsAI strategy，revealing tensions between technological innovationand institutional norms，aswell as thecomplex interplayof international cooperationand competition.

I. Endogenous Institutional Conflicts

Multiple structural tensionswithin the U.S. directly impact the achievement of its strategic AI goals.

First， the constitutional tension betweentechnological innovationandthe protection ofcitizens'rightsisbecom ing increasingly prominent. For example，the Algorithmic Accountability Act （2022）， while aiming to enhance AI transparencyand fairness，hasbeencriticized as overly burdensome regulation that could stifle innovation， highlighting the difficulty of reconciling privacy concernswithinnovationimperatives through single legislativeacts.

Second， in the realm of military， theweaponization ofAI hasraised pro found ethicalandlegal controversies， whichwas termed the“Oppenheimer Moment\". Intelligent weapons rep resentedbythe Lethal Autonomous Weapons Systems，while enhancing efficiency，also create dilemmas regardingethical legitimacyandaccountability attribution.While the 2023DoD Directive 3000.09 on “Autonomy in Weapon Systems\" strengthened human oversight，external concernspersist， reflecting the enduring tension between technological advancement and ethical governance.

Third， the concentrated control of coreAI resourcesand computing power bytech giants exacerbates the conflict between technological monopolyand innovation diffusion.With the edges in dataresources，algorithmsand digital infrastructure，the tech giants dominate everysegmentof theAIvaluechain， compressing the competitive space for SMEsand startups，ultimatelyunderminingthediversityanddynamismof the overall innovation ecosystem.

II.Realities ofInternational Coordination Challenges

The U.S.faces not only external competition but also significant challenges in coordinating with allies， managing international opinion，and adapting to emerging technologies.

First，the divergent approaches of the U.S.and the EU to data governance and privacy protection create friction.

The EU's General Data Protection Regulation，emphasizing strict data usage norms，clasheswiththeU.S.'smore flexibleand fragmented data processing system.Thisdivergence hinders transatlantic data flows，impeding collabora tive AI Ramp;D and industry cooperation.

Second，U.S.alliesinAsiaexhibit varyingprioritieson AI governance. Japan leans towardsa“human-centric\"approach，favoringcaution onalgorithmic transparencyand ethicsoversight.The ROK prioritizes employment and social equity，emphasizingAI governance forsocial stability.Thesedivergences， visiblein forumslike theG7and IPEF， complicate policy coordination and consensus-building fora“trustworthy AIalliance.\"

Third， theU.S.AI strategy faces growingcriticismglobally，especially regarding perceived “technological barriers”and“digital colonialism.”Developingcountriesexpressconcernsthat U.S.-drivenstandardsandrulescould exacerbate technological dependency and erode sovereignty. This criticism to someextentconstrainsthebreadthand depthofU.S.strategy expansion and signalstherising collectivevoice of the Global South in the technology governance discourse system.

Fourth，the integration of decentralized technologies （e.g，blockchain） with AI posesnovel challenges for traditional regulation. Their anonymity and cross-bordernaturediminishnational regulatoryefficacyandobstruct the advancementofunifiedgovernance standards.TheU.S.lacksaneffective framework to govern these“borderless technologies，”creatingastructural barrier to governance spillover.

EVOLUTIONARYTRENDSINU.S. ARTIFICIALINTELLIGENCESTRATEGY

As technological transformation acceleratesand international competition intensifies，the U.S.AI strategy is rapidlyadvancingtowardsystematic restructuring.Moving forward，the U.S.will deepen strategic adjustments acrossthreemajordimensionstechnology deployment， institutional frameworks，and security competition—aiming to consolidate its advantageous position in reshaping the global AI governance order.

I. Systematic Transformation in TechnologyandEnhancedEcosystem Coordination

TheU.S.is reshaping itsAI technologydeploymentlogic，gradually shifting froma focus onmodel training and algorithm performance toward comprehensive coordination of computing infrastructure，hardware systems，and interdisciplinary ecosystems.Underthecoordinationof the OSTP， the National Quantum Coordination Officeand the National ArtificialIntelligence Research Resource initiative，in collaborationwiththeDepartment of Energy and other agencies，are promoting the construction of“next-generation infrastructure.\" This aims to overcome computational bottlenecks foradvanced modelsand enhance infrastructure resilience and universality. Simultaneously， the DARPA is prioritizing programs such as the Fast Event-driven Neuromorphic Camera-to-Everything system and theNext-Generation Microelectronics Manufacturing initiative，driving AI hardware toward low-power，adaptive， andhigh-performancedevelopment. This shift signifies that the U.S.AI strategy has moved from singular technological breakthroughs toward systemic ecosystem planning.

Following Donald Trump's return to office，technologydeployment has placedgreater emphasisonamarketdriven and corporate investment-centered approach.In 2025，the Trump administration launched the Stargate Project， investing USD 5oo billion in collaboration with OpenAI， Oracle， andothersto buildultra-largeAIdata centers and energy infrastructure tomeetintensivetrainingdemands. Concurrently， Trump signed Executive Order 14179“Removing Barriers toAmericanLeadershipinAI，\"explic itlystating the need to “tear down unnecessaryobstaclestoAIinnovation setbythegovernment.”Thisinvolves relaxing regulatory clauses， reducing infrastructure construction restrictions，andreinforcingamarket-led， corporate-investment-drivenmodel.

II. Integration of Institutional Systemsand Rebalance ofRegulatory Mechanisms

The leap in institutional capacityisa crucial pillar supporting the stableadvancementoftheU.S.AI strategy.Currently， the federal governmentispromoting the integration ofethical norms，risk assessment，and multi-agency regulatory mechanisms. AgenciessuchastheFederalTrade Commission and the Consumer Financial Protection Bureau have progressivelyintroduced high-risk model assessmentandplatformaccountabilitymechanismsforAIsystems.In 2023，theBidenadministrationissued anexecutiveorderonAI and multiple jointagency statements， signaling the U.S.government's efforts to incorporateAI governance into routineadministrative oversight.

However， Trump'sreturn tooffice significantly altered this trajectory. TheTrumpadministrationadvocates a“deregulation-first'and‘innovationabove-all”approach，calling for a reviewofBiden-era AIpolicies.It aimstoreducefederalintervention in AI platforms， emphasizing that AI developmentshouldbemarket-led. In the meantime， Trump appointed the first White House AI and Crypto Czar to centralize AI policy oversight， reducingbureaucratichurdlesandac celerating decision-making. The Biden administration's2023ExecutiveOrder on“Safe，Secure，andTrustworthyDevelopment and Use of AI\" faces po tential suspension orreconsideration， whilethe previouslyreleased Blueprint for an AI Bill of Rightshas encounteredimplementationobstacles. This regulatory rebalancing reflects the Trump administration's strategic restructuring of AI governance： prioritizing the unleashing of technological momentum while delaying the legal codification of ethical norms.

III.Securitization of Strategic Logic and Containment Pressure AgainstChina

The focus of the U.S. AI strategy is shifting from technological innovation competitiontoward shapingglobal governance centered on security primacyand containment against China. On the one hand， through rule-making incritical areassuch asapplication programminginterfaces，dataaccess protocols，and standardizationdocumentation，theU.S.seekstoestablish global technological ecosystem dominancevia“embeddedtechnicalnorms.\" By competing in standard-setting authority，algorithmic controllability requirements，andcross-borderdata governance rules， the U.S.not only buildstechnological accessbarriersin global marketsbutalso reinforces its rule-making leadership in the future internationalorder.

On theotherhand，AIhasbeen comprehensivelyintegratedintothe U.S. national security， diplomatic， and geopolitical systems. AI is not only widelyused for countering disinformation， simulating diplomatic nego tiations，and enhancing intelligence fusion efficiency，but has also become a crucial technological tool for shaping international agendas and exporting norms. Leveraging multilateral mechanismssuchastheGPAI，IPEF，andive Eyes， theU.S.promotesthenarrative of\"trustworthy AI，\" building a digital governancealliance guided by values.

The Trump administration further institutionalized the“security-first” logic.Itannounced plans to revoke the AIDiffusion Rule implemented by the Bidenadministrationin January 2025， modifyAI chip export restrictions， and introduce a simplified global licensing system to replace thepreviousthreetiered licensing regime. This policy aimsto reduceobstaclesto U.S.innovation while strengthening national security controls over strategic technologies. Export controls against China continue to escalate， such as adding bansonNvidiaH2OandAMDMI308 chips，reinforcing model restrictions and data controls.Following the release ofDeepSeek-R1bytheChinesecompanyDeepSeek， theU.S.Navy swiftly announcedaban，and the National Security Council initiated areview— demonstrating its view of AI asa core strategicassetanditsconstructionofa multi-dimensional technological containmentsystem.

Insummary，through domestic planning，ruleexportation，andsecuritizationexpansion，theU.S.continuesto consolidate itsadvantageousposition intheglobal AI competition，demon stratingahighdegreeofsystematicity and foresight.Trump'sreturn to power has not interrupted this process but has instead manifested a stronger deregulatoryintentand nationalsecurity-driven approach in itsimplementationpath， furtherintensifying the fragmented andcompetitivenatureoftheAIstrategicsystem.Thecontinuityof the U.S. AI strategy faces challenges amid partisan transitions， yet it retains strong resource mobilization and rule-shaping capabilitiestheglobalAI governance landscape.Ultimately，whetherthe U.S.can effectivelybalance innovation， governancelegitimacy，and global responsibilitywill directlydetermine the trajectory of the global order of AI development