⚡NIMITZ TECH NEWS FLASH⚡

“AI’ve Got a Plan: America’s AI Action Plan”

Senate Committee on Commerce, Technology, and Transportation, Subcommittee on Science, Manufacturing, and Competitiveness

September 10, 2025 (recording linked here)

HEARING INFORMATION

Witnesses and Written Testimony (Linked):

• Mr. Michael Kratsios: Director, White House Office of Science and Technology Policy

HEARING HIGHLIGHTS

IN THEIR WORDS

SUMMARY OF OPENING STATEMENTS FROM THE FULL COMMITTEE AND SUBCOMMITTEE

• Subcommittee Chair Budd highlighted the subcommittee’s broad jurisdiction over job creation, economic opportunity, and competitiveness. He emphasized the importance of collaboration with Ranking Member Baldwin and praised the Trump administration’s leadership on innovation, citing supersonic flight as an example. He stated that AI had transformative potential similar to the internet and stressed that policy should focus on maximizing economic opportunity for everyday Americans, not just Silicon Valley billionaires. He warned against China’s state-backed model and advocated for a private sector–led approach, regulatory certainty, permitting reform, and strong domestic manufacturing to ensure U.S. leadership in AI globally.

• Subcommittee Ranking Member Baldwin underscored the promise of AI to modernize the grid, improve severe weather forecasting, drive agricultural innovation, and support scientific and medical breakthroughs. She stressed that AI must be developed responsibly, with clear guardrails to protect rights and innovation. She criticized the Trump administration for cutting billions from science and research funding, arguing that such actions undercut U.S. competitiveness and undermine talent development. She concluded by emphasizing the need to support researchers and educators if America is to lead in AI.

• Full Committee Ranking Member Cantwell thanked colleagues for bipartisan efforts to advance AI legislation and praised the administration’s executive order for aligning with priorities such as education, training, and capacity building. She drew attention to the Pacific Northwest’s leadership in data centers and affordable energy, linking this to the importance of energy capacity for AI leadership. She also raised concerns about China expanding influence in the Middle East and urged that U.S. technology exports set global standards to prevent misuse by bad actors. She expressed support for continued innovation in next-generation energy and supply chains, emphasizing the need for the U.S. to build a globally adopted AI framework.

• Full Committee Chair Cruz stated that regulating AI was among the most critical policy challenges of the time, describing AI as transformative with potential to raise living standards and expand American values. He warned that the U.S. was in an AI race with China and praised the Trump administration’s AI Action Plan for enabling innovation. He introduced his legislative framework for AI, focusing on innovation, free speech, regulatory consistency, preventing misuse, and defending human dignity. He announced the “Sandbox Act” to create regulatory flexibility while maintaining accountability, stressing that American leadership was essential to ensure AI reflected U.S. values rather than authoritarian ones.

SUMMARY OF WITNESS STATEMENT

• Mr. Kratsios explained that the President’s AI Action plan built on President Trump’s 2018 initiative and reflected a renewed commitment to American AI leadership after the U.S. lead had narrowed by 2024. He emphasized the administration’s vision of a “Golden Age” of innovation, characterized by scientific rigor, private industry partnerships, and university research, and stated that the executive and legislative branches must work together to preserve U.S. preeminence.

  He outlined the three pillars of the AI Action Plan—innovation, infrastructure, and international partnerships—each supported by executive orders. These included preventing “woke AI” in federal use, accelerating permitting for data centers, and promoting the export of American AI technology. He noted OSTP’s role in supporting Commerce Department requests on export frameworks and stressed that U.S. global leadership depended on international adoption of American AI systems.

  Mr. Kratsios called for congressional collaboration to ensure regulatory clarity, including regulatory sandboxes for product testing and federal preemption of conflicting state rules. He highlighted the need for permitting reform and investment in physical infrastructure such as data centers, minerals, and energy capacity. He also stressed that AI adoption must begin domestically, from self-driving cars to legislative tools, and should benefit all Americans through workforce training, education, and small business support.

SUMMARY OF KEY Q&A

• Chair Budd asked for an update on early implementation of the AI Action Plan. Mr. Kratsios said progress had been made on the export package, education initiatives, and regulatory reviews. Budd asked about the American AI technology stack. Kratsios said it consisted of chips, algorithms, and applications, and stressed the importance of promoting U.S. technology globally before China caught up. Budd asked about the downside if the U.S. stack was not adopted worldwide. Kratsios warned that adversaries’ models and chips would dominate instead, threatening U.S. interests. Budd asked whether private companies or the government should lead international adoption. Kratsios said both would work together, with the government assisting industry in outreach and access to foreign markets.

• Ranking Member Baldwin raised concerns about data centers draining Great Lakes water. Mr. Kratsios said the administration was committed to the highest standards of clean air and water. Baldwin asked how expedited Clean Water Act processes would protect resources. Kratsios said regulatory changes would undergo notice and comment to maintain high standards. Baldwin asked about AI applications for farmers. Kratsios said precision agriculture showed the most promise and could increasingly benefit smaller farms. Baldwin asked about AI in severe weather forecasting. Kratsios deferred to NOAA but said AI would strengthen U.S. forecasting capacity. Baldwin asked about modernizing the electric grid. Kratsios said AI could improve load balancing despite the grid’s complexity and noted DOE’s commitment to the effort.

• Chair Cruz asked if the U.S. could beat China in AI without congressional action. Mr. Kratsios said executive and legislative collaboration was essential.

  Cruz asked why regulatory sandboxes were helpful. Kratsios said they allowed innovators to test technology safely and inform regulators. Cruz asked if he supported Congress establishing AI sandboxes. Kratsios confirmed his support and eagerness to collaborate. Cruz asked why state AI rules were harmful.Kratsios said a patchwork of regulations hindered innovation and advantaged large firms, so federal preemption was needed. Cruz cited Colorado’s reporting requirements and asked about the risks. Kratsios said such regulations were anti-innovation and reinforced the case for preemption. Cruz asked how to push back on foreign censorship rules. Kratsios said international standard-setting bodies should advance free speech principles.

• Ranking Member Cantwell thanked Mr. Kratsios for focusing on exports, data centers, and legislation, and highlighted bipartisan bills on NIST standards, AI research resources, and AI education that had stalled but could have advanced U.S. leadership. Mr. Kratsios said recent deals in Saudi Arabia and the UAE showed U.S. commitment to exporting chips and making America the partner of choice in the region. Cantwell asked if a “technology NATO” could be created with allies. Kratsios said the export program provided an opportunity to build a trusted network with partners and to design a modular system that incorporated allied technologies. Cantwell asked about sandboxes and whether they applied to specific applications. Kratsios said sandboxes were useful where older laws blocked AI use cases, but final implementation would rest with agencies. Cantwell clarified that he did not see himself as an AI “czar,” and Kratsios confirmed it would be agency-led. Cantwell questioned reports that the Secretary of Commerce suggested taking 50% of university startup revenues. Kratsios said he was not aware of the comment but stressed OSTP had long supported federal R&D without revenue seizure. Cantwell reiterated opposition to such an idea and thanked him for collaboration on bipartisan AI bills.

• Sen. Schmitt raised concerns about bias in large language models, citing examples from ChatGPT and the use of Wikipedia rankings. Mr. Kratsios said this was why President Trump signed an executive order against “woke AI” and directed OMB to ensure government-procured models were truth-seeking and accurate, with penalties for violations. Schmitt said a competitive marketplace was needed to prevent monopolies and promote truthful models. Kratsios agreed, criticized the prior administration’s approach, and said the action plan emphasized open-source models and procurement standards to foster accuracy.

• Sen. Blunt Rochester said she had seen errors in ChatGPT about herself and stressed the issue was about models being smart or dumb, not woke or sleepy. Mr. Kratsios agreed that outputs reflected inputs. Blunt Rochester highlighted Delaware’s leadership in AI skills training and a state-level sandbox, while noting skepticism about OSTP’s role in a federal sandbox. She asked how the AI Action Plan would build on programs like Manufacturing USA. Kratsios said the plan prioritized infrastructure build-out, workforce training, and retraining to support advanced manufacturing jobs. Blunt Rochester asked about leveraging institutions like the National Institute for Innovation in Manufacturing Biopharmaceuticals (NIMBL) for biosecurity. Kratsios said AI could be used for bioscience breakthroughs and automated labs, creating novel compounds for national benefit. Blunt Rochester cautioned against cuts to NSF and STEM programs, urging protection of workforce and innovation capacity. She thanked the witnesses and yielded back.

• Chair Blackburn noted bipartisan bills on standards, precision agriculture, an innovation-in-agriculture bill, and a quantum sandbox, asked that OSTP submit its list of AI-inhibiting regulations to the committee, and pressed for a national online privacy standard. Kratsios agreed to submit the list, affirmed that online privacy was critical, and said he was eager to work with the committee.

  Blackburn asked how national labs should collaborate on the American Science Cloud. Kratsios said interoperability and data transfer across lab and private supercomputing infrastructure should be a primary objective. Blackburn warned about LLM training on patented and copyrighted content and asked how the administration would firewall such material while allowing current-events conversation. Kratsios was invited by the Chair to address the issue later, and the hearing moved on.

• Sen. Peters argued that weak data governance made rapid AI adoption risky and cited the Social Security Administration’s CDO resignation tied to DOGE endangering data. Mr. Kratsios said he was not familiar with that example but emphasized that data protection was critically important across agencies.

  Peters pressed who would decide which state AI laws were “prudent” versus “unduly restrictive” for federal funding eligibility. Kratsios said relevant funding agencies and their secretaries would make those determinations. Peters asked whether Michigan would lose funds for laws against AI-enabled sexual exploitation and political deepfakes and cited reports of “Grok” producing hate speech despite OMB guidance. Kratsios said the administration required government-procured AI to be truth-seeking and accurate under the executive order and took any model bias seriously.

• Sen. Moreno asked whether government was built for innovation and whether private-sector leadership was essential to compete with China. Mr. Kratsios agreed, saying that philosophy underpinned the plan. Moreno emphasized U.S. dominance in chips, models, and applications and highlighted Intel’s Ohio fab.Kratsios agreed on the full stack, said the U.S. must both design and fabricate chips domestically, and noted the administration’s commitment to Intel. Moreno stressed abundant, reliable, co-located energy for AI data centers and asked about AI’s energy intensity versus search. Kratsios agreed that energy co-location was important and said industry reported AI workloads were far more energy-intensive and growing rapidly. Moreno asked about developing top talent and modernizing government through private contracts that include small firms. Kratsios pointed to pillar-two workforce programs across Labor, Education, and Commerce, cited GSA’s FedRAMP improvements and DoD’s Tradewinds to speed AI procurement, and agreed that small-business access mattered.

• Sen. Rosen warned against cutting workforce-building agencies and asked Mr. Kratsios to commit that agencies would not use AI systems that promoted antisemitic conspiracy theories or hate speech. Kratsios said the administration would execute the executive order requiring government-procured models to be truth-seeking and accurate, and agreed that the behavior she described would not meet that standard. Rosen asked about accelerating AI fiber infrastructure and the roles of FCC and NTIA. Kratsios said fiber was a critical interconnect for AI and noted NTIA and Commerce were taking it seriously. Rosen asked whether eliminating broadband programs undermined future connectivity. Kratsios said fiber was one option among several and that NTIA evaluated the most economical approaches. Rosen asked him to affirm funding for connectivity. Mr. Kratsios agreed that connectivity was critically important.

• Sen. Markey asked if he knew how much household electricity bills would rise due to data-center expansion and whether it was appropriate to push growth that raised costs. Mr. Kratsios said he did not know the figure and asserted the administration was committed to expanding power generation and lowering energy costs. Markey argued the administration was stifling renewables while promoting data centers and asked if AI procurement rules against “top-down ideological bias” would disqualify models trained to adopt a political viewpoint. Kratsios said guidance was pending but a model that was not truth-seeking and accurate would violate the executive order. Markey cited Grok’s statement that it was trained to “appeal to the right” and asked if that should disqualify it from federal contracts. Kratsios said models failing the truth-seeking and accuracy standard would be subject to procurement restrictions.

• Sen. Young asked about the importance of cloud labs for biotech R&D and their impact on innovation speed. Mr. Kratsios said automated, AI-driven experimentation would dramatically accelerate discovery and noted NSF activity on cloud labs. Young asked about risks from fragmented standards at home and abroad. Kratsios said the U.S. must lead on global AI standards—especially for model evaluation—given rivals’ efforts to set competing rules.

  Young said he would reintroduce the Future of AI Innovation Act and asked for collaboration. Kratsios agreed to work with him on the legislation.

• Sen. Hickenlooper asked if principles such as risk assessments, transparency, disclosure, privacy protections, and standards development should form the basis of a federal AI law. Mr. Kratsios said broad regulations were not the right approach and argued that sector-specific, use-case regulation was preferable to avoid Europe’s mistakes. Hickenlooper suggested some issues, like identifying AI-generated content, should be general. Kratsios agreed, saying research on detection was important to fund. Hickenlooper asked about independent evaluations of AI models and referenced his bipartisan VALIDATE Act. Kratsios said evaluation science was underdeveloped, and his priority for NIST was advancing metrology so agencies could use scientifically backed methods. Hickenlooper asked about workforce development and apprenticeships. Kratsios said the President and the Labor Department were strongly committed, with a goal of one million new apprenticeships this term.

• Ranking Member Klobuchar described the risks of deepfakes and highlighted her Take It Down Act and No Fakes Act, asking if unauthorized likeness replication should be restricted. Mr. Kratsios agreed and said such protections should become law. Klobuchar pressed for guardrails distinguishing between protected parody and harmful deepfakes and asked if he would work with her and Sen. Thune on federal standards. Kratsios said he would. Klobuchar referenced whistleblower testimony that Meta suppressed safety research and asked for his commitment to address AI chatbot harms. Kratsios said he would work with Congress and cited the administration’s AI education task force as part of its effort to equip youth with awareness of AI’s limitations.

ADD TO THE NIMITZ NETWORK

Know someone else who would enjoy our updates? Feel free to forward them this email and have them subscribe here.