Today, the Biden Administration released a National Security Memorandum (NSM) on artificial intelligence. At a high policy level, the NSM directs the executive branch to take steps to, “(1) ensure that the United States leads the world’s development of safe, secure, and trustworthy AI; (2) harness cutting-edge AI technologies to advance the U.S. Government’s national security mission; and (3) advance international consensus and governance around AI.” According to the White House, the NSM’s, “fundamental premise is that advances at the frontier of AI will have significant implications for national security and foreign policy in the near future.”
This memo has been long expected, as it was mentioned at the end of the AI Executive Order, which was released at the end of October 2023.
The NSM takes several steps to advance the use and adoption of AI for national security purposes. From directing improvements to the security and diversity of semiconductor chips supply chain; to “formally designating the AI Safety Institute” at NIST as the primary point of between US industry and the federal government; to doubling down on the NAIRR pilot; to directing the creation of a “Framework to Advance AI Governance and Risk Management in National Security;” to directing the US Government to collaborate with allies and partner nations to, “establish a stable, responsible, and rights-respecting governance framework,” to use AI in ways that adhere to international law and respect human rights. More details can be read in the Fact Sheet released by the White House.
This is a good step to continue the adoption of AI in the operations of the federal government, as well as to do so in a socially responsible way. However, more remains to be done. CRA will monitor the actions taken around AI at the federal level and we will continue to advocate that fundamental AI research, as well as research into its use and applications, must be a key part of any national strategy around this technology.
Update 9/26/24: The Senate approved the CR last night and sent the bill to President Biden to be signed into law.
Original Post: With less than a week till the beginning of Fiscal Year 2025 (October 1st), Congress is rushing to pass a continuing resolution (or CR) to avoid a government shut. Over the weekend, Congressional leaders announced a bicameral, bipartisan compromise to pass a relatively clean CR that would extend funding authority, at FY24 levels, to December 20th. This is likely the best-case scenario given the situation, as House Speaker Johnson (R-LA) originally floated a March 2025 CR with a politically toxic voting amendment attached to it. A clean CR, which does not contain politically contentious policy provisions or funding changes, avoids a messy political fight and the possibility of a government shut down until after the election.
However, confidence that Congress will finalize FY25 by December 20th would be misplaced. As we have said repeatedly during the budget process, the outcome of the November elections will directly impact how FY25 is closed out. If the political calculus is relatively unchanged after the election (meaning a split in control of Congress and Democrats retain the White House), it is likely that this Congress will want to close out FY25 to give the 119th Congress a clean slate. However, if either side gains a clear advantage in the election, the likely outcome will be punting the matter into 2025, where that side will have more say over the funding details. We will have to wait for events to play out.
On September 11th, the House Science, Space, and Technology Committee passed nine pieces of legislation covering a diverse set of topics around artificial intelligence (AI). The bills are now sent to the full House of Representatives for consideration and potentially passage into law.
NAIRR, a cyberinfrastructure resources proposed by a Congressionally established task force of the same name, was started up as a pilot program by NSF at the beginning of 2024. The goal of NAIRR, which will be run by NSF and overseen by an interagency steering committee, is to provide, “free or low-cost access to datasets and computing resources for development of AI workflows,” helping to democratize the development and use of artificial intelligence. The bill version would authorize NAIRR to receive $430 million a year for five years (FY2025 to FY2030).
Of all the bills the Science Committee moved on the 11th, this is the most likely to become law. That is because there is a Senate version that was passed by the Senate Commerce Committee over the summer and it enjoys wide, bipartisan support in both chambers of Congress. However, expectations on that happening should be kept in check, as there are few legislative working days left in the year and Congress does not have many legislative vehicles to move this to passage. If it were to move, it would likely do so as an amendment in the annual defense policy bill. There are also rumors of an AI supplemental funding bill during the lame duck session of Congress; however, this author would caution that until we know the outcome of the November elections, such an effort is unlikely.
H.R. 9402, the NSF AI Education Act of 2024
This bill support NSF’s education and professional develop mission related to AI. First, it allows NSF to support scholarships and fellowships in AI, and specifically includes community colleges for support. It also directs NSF to support professional development for students, teachers, faculty, and industry professionals. This includes supplements to students and faculty to attain skills, training, or education in AI, as well as fellowships for industry and school professionals.
Next, this legislation establishes a “Centers of AI Excellence” program at NSF. The agency is directed to run the program in coordination with NIST’s Regional Technology Hubs, while also leveraging the NSF Engines program and other NSF efforts it deems necessary and useful. The establishment of such a center is to, “enhance educational outcomes and drive workforce development by integrating artificial intelligence into teaching, learning, and community engagement.” The bill’s language specifically includes community colleges and “area center and technical educational schools” to the list of groups who can take part in the Center program.
The legislation also directs NSF to make awards to promote research regarding teaching models, and materials for AI and its integration into classrooms, teaching, and learning for Pre-K through grade 12 students who are from low-income, rural, or Tribal populations. Finally, the legislation incorporates AI skills development into the National STEM Teacher Corp.
H.R. 9211, the LIFT AI Act
The Lift AI Act would allow NSF to make awards to institutions of higher education or nonprofit organizations to support research activities to develop educational curricula and evaluation methods for AI literacy at the K-12 level. It also allows NSF to carry out these activities through new or existing funding; it does not provide additional funding authorizations.
There is also a sense of Congress clause that talks about the importance of AI literacy in K-12 for the nation’s future workforce and how it underpins the country’s economic and national security. Such clauses don’t have much legal strength, but they provide a useful rhetorical point for the research agencies to say “Congress deems this subject important and that’s why we’re funding these activities.”
H.R. 9403, the Expanding AI Voices Act
This bill directs NSF to broaden participation and capacity in AI research, education, and workforce development. It does this by competitive awards to institutions that are not one of the top 100 institutions (as determined by Federal R&D expenditures during the 3-year period prior to the year of the award), HBCUs, MSIs, tribal colleges or universities, or a consortia of any of the these entities.
An eligible institution may use the funds to carry out:
Development or expansion of research programs in AI and related disciplines.
Faculty recruitment and professional development in AI.
Bridge programs focus on preparing post-baccalaureate students for grad programs in AI.
Provide or broker access to research resources, including computer resources, networking, data facilities, and software engineering support for AI R&D.
Community building activities to foster mutually beneficial public-private collaboration with Federal research agencies, industry, Federal laboratories, academia, and nonprofit organizations.
Development and hosting of intra- or inter-institutional workshops to broaden workforce participation in AI.
Activities to integrate ethical and responsible practices and principles into education programs in AI
Other activities necessary to build research capacity, education pathways, and workforce development pathways in AI.
NSF is then directed, when performing outreach to the community about this program, to take into account all regions of the country, and to especially consider people from underserved communities and groups historically underrepresented in STEM. No additional funding authorization is provided.
H.R. 9215, the Workforce for AI Trust Act
This bill is meant to, “facilitate a workforce of trained experts to build trustworthy AI systems.” The legislative language is split between sections for NSF and NIST.
For NSF, it allows the agency to support graduate and postdoc research fellowships across disciplines. The language includes humanities and social sciences for fields that should be included. The fellowships are meant for the, “integration of ethical and responsible practices and principles into the design, development, training, deployment, evaluation, and understanding of artificial intelligence systems.” The bill also directs NSF to make awards for the development and hosting of intra and inter-institutional workshops on integrating perspectives and skills from multiple disciplines toward the deployment, evaluation, and understanding of AI systems. Finally, it directs NSF to integrate these perspectives into the agency’s peer review process.
For the NIST section of the bill, it amends NIST’s AI mission to have the agency support education and workforce development activities to expand the AI workforce. This includes careers related to helping organizations govern, map, measure, and manage AI related risks, including testing, evaluation, verification, and validation of AI systems.
H.R. 9466, the AI Development Practices Act
This bill directs NIST to “catalog and evaluate emerging practices and norms for communicating certain characteristics of artificial intelligence systems, including relating to transparency, robustness, resilience, security, safety, and usability, and for other purposes.” This is perhaps the most esoteric bill that the committee considered, but it covers an essential function of the federal agency whose mission is to assemble standards for industry. The language makes clear that any guidance that NIST develops must remain voluntary.
H.R. 9497, the AI Advancement and Reliability Act
This bill establishes a “Center for AI Advancement and Reliability” at NIST in order to ensure US leadership in, “research, development, and evaluation of the reliability, robustness, resilience, security, and safety of artificial intelligence systems.” The center is to coordinate with NSF, OSTP, DOE, DOD, DHS, and other departmental secretaries or agencies as considers appropriate. The bill also directs NIST to establish a consortium of stakeholders from academic, industry and civil society. The center is authorized to receive $10 million for Fiscal Year 2025.
H.R. 9197, the Small Business Artificial Intelligence Advancement Act
This is a small business assistance bill that directs NIST to provide guidance to help such entities utilize advances in the AI marketplace.
H.R. 9194, the Nucleic Acid Screening for Biosecurity Act
Finally, the shortest bill the Science Committee considered, it’s a biology sciences bill that touches on the potential impact AI could have on nucleic acid screening sciences.
Today six leading organizations — AAAI, ACM, CRA, IEEE-USA, SIAM, and USENIX — representing more than 400,000 people in computing, information technology, science, and innovation across US industry, academia, and government, joined together to call on Congressional leaders of both parties to fully fund the research agencies contained in the CHIPS and Science Act of 2022:
September 11, 2024
The Honorable Mike Johnson
Speaker of the House
United States House of Representatives
Washington, DC 20515
The Honorable Mitch McConnell
Republican Leader
United States Senate
Washington, DC 20515
The Honorable Chuck Schumer
Majority Leader
United States Senate
Washington, DC 20515
The Honorable Hakeem Jeffries
Democratic Leader
United States House of Representatives
Washington, DC 20515
Dear Speaker Johnson, Leader Schumer, Leader McConnell, and Leader Jeffries:
As six leading organizations representing more than 400,000 people in computing, information technology, science, and innovation across US industry, academia, and government, we urge Congress to appropriate the vital funding levels authorized in the CHIPS and Science Act of 2022 for the National Science Foundation (NSF), the Department of Energy’s (DOE) Office of Science, the National Institute of Standards & Technology (NIST), and other critical technology and innovation-driven agencies.
For several years and across both parties, Congress has rightly prioritized U.S. leadership in the global future of critical technologies and industries. Investments in fundamental R&D across the federal government foster innovative breakthroughs, drive job growth, and ensure the country’s national and economic security amidst growing global competition.
The U.S. is the current world leader in R&D of critical and emerging technologies because of our dynamic research ecosystem, a key component of which is the federal investment in fundamental research. We have seen time and time again, from the Internet to exascale computing to artificial intelligence, that breakthroughs supported by the federal research agencies pay huge dividends to the country and help launch new sectors of the economy. By one measure, the return on investment of non-defense government R&D is between 150 and 300 percent. All Americans should be proud of that record of accomplishment and the return on their investment.
But the nation’s commitment to fundamental research, particularly in the computing and IT fields, has been slipping. Significant cuts to the budgets of NSF, NIST, and DOE in FY2024 will have a serious impact on the future of U.S. leadership in technological innovation. Fewer good ideas and new research visions will see investments, fewer graduate students will be produced, and the innovation economy driven by what the National Science Board has described as the “extraordinarily productive interplay between academia, government, and industry” will move forward more slowly – at a time when our global adversaries are increasing their investments. A nation that is striving to lead the world in artificial intelligence, quantum computing, high performance computing, and related fields cannot do that with cuts or flat funding at NSF, DOE, NIST, and the other research agencies.
Congress should fund these agencies at the highest amount possible in the Fiscal Year 2025 budget and look at opportunities for supplemental funding, such as in artificial intelligence legislation, to meet our emerging technology and computing research, workforce, and infrastructure needs. We are not alone in calling on the nation’s leaders to prioritize the commitments made to the country’s research efforts at NSF, DOE, NIST, and the other federal research agencies, at the levels authorized in the CHIPS and Science Act of 2022 – the National Science Board, the Science and Technology Action Committee, and chief technology officers from across industry have recently made similar calls. This is not a partisan issue; it must be an American priority. We cannot risk missing this moment and losing our hard-fought position as the world leader in scientific research to our competitors.
Association for the Advancement of Artificial Intelligence (AAAI) aaai.org
At the beginning of August, the Senate Appropriations Committee completed its work on its slate of Fiscal Year 2025 funding legislation. In their last bundle of bills was the chamber’s version of the Defense appropriations legislation, which handles funding for the Department of Defense. Unfortunately, the Senate’s plans are generally bad for the defense research accounts, calling for cuts or flat funding for basic, applied, and advanced technology development research areas.
As a refresher, DOD’s Science and Technology program is made up of three accounts: 6.1 (basic research), 6.2 (applied research), and 6.3 (advanced technology development). These accounts are themselves made up of individual accounts for each of the three services (Army, Navy, and Air Force), as well as a Defense Wide account. DARPA, or the Defense Advanced Research Projects Agency, is a section under the Defense Wide account.
Getting into the details of the Senate’s proposal, Basic Research (6.1), which is the main Defense Department supporter of fundamental research at US universities, would receive a cut of 4.2 percent compared to its FY24 levels. The account goes from $2.63 billion in FY24 to $2.52 billion for FY25, a decrease of $110 million. A good amount of this decrease is due to reductions in the different services’ “University Research Initiative” subaccounts: the Army’s subaccount decreases by 8.8 percent, the Navy’s by -6.7 percent, and the Air Force’s by -20 percent. Only the Space Force received an increase (2.3 percent). The only other bright spot is the Defense-Wide account, which received an overall increase of about 2.0 percent.
The Applied Research (6.2) account fairs the worst among the defense accounts, receiving a significant cut in funding under the Senate’s mark. The full account would see a 14.1 percent decrease compared to last year’s budget, going from $7.60 billion in FY24 to $6.53 billion under the Senate’s plan, a reduction of $1.07 billion.
The Advanced Technology Development (6.3) account would be flat funded, receiving a slight cut under the Senate’s budget framework. It would go from $11.29 billion in FY24 to $11.20 billion in FY24, a decrease of $90 million (or -0.8 percent).
Finally, DARPA would be the one bright spot for defense research in the Senate’s funding mark. The agency would receive a 17 percent increase and go from $4.12 billion in FY24 to $4.84 billion in FY25, an increase of $720 million.
FY24
FY25 PBR
FY25 House
FY25 Senate
$ Change
% Change
DOD 6.1
$2.63B
$2.45B
$2.52B
$2.52B
-$110M
-4.2%
DOD 6.2
$7.60B
$5.80B
$6.61B
$6.53B
-$1.07B
-14.1%
DOD 6.3
$11.29B
$9.00B
$9.86B
$11.20B
-$90M
-0.8%
DARPA
$4.12B
$4.37B
$4.21B
$4.84B
-$720M
+17%
This is yet another example of a poor Presidential request creating problems further down the budgetary path. When one compares the Senate numbers against the President’s request from March, the 6.1, 6.2, and 6.3 accounts would receive a 2.9, 12.7, and 25 percent increase respectively. Put another way, the Senate looked after these accounts by giving them more than DOD requested, even though they are less than the accounts received a year before. While the Defense Department is in a difficult position with its overall budget, saying back in March that their planning comes from a zero-growth mindset, this is still poor support for the nation’s defense research efforts.
It’s hard to say where a final defense research budget will ultimately settle. The House and Senate are not far apart with their budget plans, which could make final numbers easy to agree. Unfortunately, with both chambers calling for cuts or flat funding, the likely final numbers could be poor. However, it is clear that nothing will be decided by the beginning of the new Fiscal year (October 1st). And as we have said about all the other research agencies, a final FY25 budget won’t be decided until after the November elections. In fact, depending on the outcome of the election, a final resolution for FY25 may slip into the next calendar year. CRA will continue to monitor the situation and report on any new develops; please keep checking back for the latest news.
Continuing our coverage of the Fiscal Year 2025 (FY25) federal budget process, we turn to the Senate Appropriations Committee’s Energy and Water bill. This bill contains the budgets for the Department of Energy’s Office of Science (DOE SC) and ARPA-E. The Senate Appropriators supported the President’s requested budget from March and provided generous funding increases for the Office of Science and ASCR. This is contrast to the lower funding levels the House appropriators recommended with their budget plan.
The Senate bill proposes a 4.7 percent increase for the Office of Science over FY24 enacted levels, bringing the agency’s budget to $8.60 billion for FY25 (an increase of $360 million). Within the Office of Science, the Advanced Scientific Computing Research (ASCR) program, which houses the majority of the computing research at DOE, would see a substantial increase of 12.7 percent – going from $1.02 billion in FY24 to $1.15 billion for FY25. These numbers are substantially the same as what the Department requested for their budgets. The Senate appropriators made a point of calling out scientific discovery and Chips & Science Act implementation as priorities in their bill, saying it, “continues to advance the highest priorities in materials research, high-performance computing, artificial intelligence, biology, and clean energy research to maintain and strengthen our global competitiveness.”
In the committee’s report, the Senate appropriators continued their support for the agency’s efforts in artificial intelligence and quantum information sciences. For AI specially, they are recommending a new line item of $100 million for artificial intelligence and machine learning efforts, and voicing support for the Department’s new Frontiers in Artificial Intelligence for Science, Security, and Technology [FASST] initiative. The appropriators also recommend no less than $160 million for AI efforts across the Office of Science, with a leadership role for ASCR, and encouraged DOE, “to apply those capabilities to the Office of Science’s mission with a focus on accelerating scientific discovery.” Within ASCR, support is voiced for the program’s Mathematical, Computational and Computer Sciences research efforts. Finally, the Senate directs DOE to establish, “a hybrid High-Performance Computing [HPC]/Quantum Computing Pathfinder Program at a Leadership Computing Facility,” in order to allow the LCF to, “to acquire an on-premise quantum computer…to begin to study how to effectively interface and integrate quantum processing units with traditional HPC resources.”
The Senate appropriators also voiced their support for the Office of Science’s engagement plans with HBCUs and MSIs, with particular praise for the agency’s RENEW and FAIR programs. This is in contrast to the House’s plans which eliminated funding for the programs. The Senate budget mark also support’s DOE’s EPSCoR program, recommending $35 million.
Finally, the Advanced Research Projects Agency – Energy, or ARPA-E, would receive a light cut of a million dollar, or a reduction of 0.2 percent compared to FY24 enacted levels. The agency would go from $460 million in FY24 to $459 million under the Senate’s FY25 plans. That is above both the President’s request and the House’s mark.
FY24 Final
FY25 PBR
FY25 House
FY25 Senate
$ Change
% Change
DOE SC Total
$8.24B
$8.60B
$8.39B
$8.60B
+$360M
+4.7%
ASCR
$1.02B
$1.15B
$1.11B
$1.15B
+$130M
+12.7%
ARPA-E
$460M
$450M
$450M
$459M
-$1M
-0.2%
The Senate Appropriations Committee unanimously approved their Energy and Water legislation on August 1st; the bill now heads to the full chamber for consideration. As we have mentioned before, we are not expecting the Fiscal Year 2025 budget to be completed until after the November election. And the results of that election could mean that finalizing FY25 is punted until the new Congress in the next calendar year. We will have to wait and see what happens. CRA will continue to monitor the situation and report on any new develops; please keep checking back for the latest news.
In our ongoing series looking at the Fiscal Year 2025 budget, we turn to the Senate Appropriations Committee and their Commerce, Justice, Science legislation. This bill is of most concern to the computing research community, as it contains the budgets for the National Science Foundation (NSF), the National Institute of Standards and Technology (NIST), and NASA. The Senate numbers are generally better than what the House appropriators recommended in July, while not as good as what the Biden Administration requested in their funding request. This was accomplished by expanding the allocation of funds that are available for this bill, a departure from the House approach, and prioritizing funding for implementing the Chips and Science Act activities around AI and semiconductor development.
Getting into the numbers, under the Senate’s plan NSF would receive a 5.4 percent increase, going from $9.06 billion in FY24, to $9.55 billion in FY25, an increase of $490 million. While this is a good number, it is still below the levels set in the Fiscal Year 2023 budget, and what the Biden Administration requested in March.
The increase to NSF is evenly distributed throughout the agency. The Research and Related Activities (R&RA) account, which houses the agency’s research portfolio, would receive $7.53 billion for FY25, which is an increase of 4.9 percent or a plus up of $350 million. The Directorate for STEM Education (EDU) would receive $1.23 billion, which is an increase of $60 million from FY24, or +5.1 percent. These numbers are generally better than what the House appropriators recommend under their funding legislation, with the exception of RRA, which the House prioritized and provided slightly more funds.
The policy provisions contained in the committee’s report had several topics related to computing issues in the section devoted to RRA. Artificial intelligence features prominently, and the Senate appropriators provide “not less than” the FY24 funding levels for AI research. Additionally, the committee, “encourages NSF to fund meritorious research and develop technical methods and techniques to improve the transparency, interpretability, and explainability of AI to better understand why and how models arrive at their decisions, recommendations, and other outputs.” Also, the National AI Research Resource (NAIRR) was provided with $30 million in funding. With regard to quantum research, the committee fully funds the President’s requested budget and, “encourages NSF to partner with institutions of higher education, industry, and other Federal agencies in order to develop the next generation of quantum computing workforce,” using grants and interdisciplinary research initiatives geared toward workforce development. The committee continues to support the TIP Directorate and reminded NSF that the success of the directorate, “will be enhanced through investing in the necessary foundational basic research provided by scientific disciplines across the research spectrum.” Within that, the Senate provides $200 million for the NSF Engines program and directs the agency to award at least 20 percent of the program’s funds in EPSCoR states. In a related area, the Senate appropriators directed NSF, “to the maximum extent practicable, 16.5 percent of NSF research funding and 20 percent of scholarship funding” should go to EPSCoR States; this is in line with the policy set out in the Chips and Science Act. Finally, the committee voiced its support for NSF’s research security efforts and encourages the agency to continue to work with the community on implementation.
With regard to provisions for the STEM Education Directorate, the Senate appropriators are setting up a clash with their House counterparts on several topics. For example, the Senate Committee “maintains its strong support for NSF’s informal science education program,” and provides $70 million; the House appropriators zeroed out the budget for this program under their plan. The Senate appropriators also make a point of voicing its support for NSF’s broadening participation efforts.
FY24
FY25 PBR
FY25 House
FY25 Senate
$ Change
% Change
NSF Total
$9.06B
$10.20B
$9.26B
$9.55B
+$490M
+5.4%
R&RA
$7.18B
$8.05B
$7.55B
$7.53B
+$350M
+4.9%
EDU
$1.17B
$1.30B
$1.00B
$1.23B
+$60M
+5.1%
The National Institute of Standards and Technology (NIST) would continue to have a complicated budget under the Senate’s FY25 funding plan. At the topline NIST would receive $1.53 billion for FY25, a 4.8 percent increase, or $70 million more than it received in FY24 ($1.46 billion). However, the Science and Technical Research and Services (STRS), would receive $1.06 billion, a decrease of $20 million, or -1.9 percent, compared to FY24. STRS contains the majority of the agency’s research portfolio. Both of those numbers are better than what the Administration requested in March and the House approved in July.
Regular readers will recall that NIST’s budget situation has become complicated over the last several years due to the large number of earmarks Congress has approved in its budget. The Senate’s proposed funding plan is no different. This makes any apples-to-apples comparison of NIST’s budget difficult. When the earmarks are removed, STRS would receive an increase of over 13 percent compared to FY24 levels (these numbers are tracked by AIP FYI).
In terms of policy provisions, the committee provides up to the requested levels to, “advance AI research, standards, testing, and the U.S. AI Safety Institute [USAISI],” at the agency. The Senate appropriators were also highly complementary of the creation of the USAISI, and that NIST sought public input from the community on several topics and directs the agency to continue these efforts. With regard to quantum information sciences research, the committee provides not less than the FY24 levels for these efforts. And the committee’s report contains several other computing related topics such as cybersecurity, digital twin technology, and firmware resiliency.
FY24
FY25 PBR
FY25 House
FY25 Senate
$ Change
% Change
NIST Total
$1.46B
$1.50B
$1.42B
$1.53B
+$70M
+4.8%
STRS
$1.08B
$975M
$1.00B
$1.06B
-$20M
-1.9%
Finally, NASA receives a slight increase under the Senate’s plans. At the topline, NASA would receive $25.4 billion for FY25, an increase of $550 million from FY24 levels (+2.0 percent). The agency’s Science account would receive a slight better increase than the topline, receiving $7.57 billion, an increase of 3.3 percent, or +$240 million, over FY24 ($7.33 billion).
FY24
FY25 PBR
FY25 House
FY25 Senate
$ Change
% Change
NASA Total
$24.9B
$25.4B
$25.2B
$25.4B
+$550M
+2.0%
Science
$7.33B
$7.57B
$7.33B
$7.57B
+$240M
+3.3%
The Senate Appropriations Committee passed the CJS bill with bipartisan support on July 25th; it is unclear when or if it will be considered by the full Senate chamber. With the House CJS legislation passed in July, technically negotiations to finalize these budgets can begin between the two chambers. However, no one in Washington believes that FY25 will be completed until after the November general election. And the outcome of that election will heavily influence how things are handled; if one party or the other comes out with a clear advantage, final consideration of the budget will likely to be punted into the 2025 calendar year. We still have a long way until Fiscal Year 2025 is completed. CRA is continuing to monitor the situation for any developments, so please check back for updates.
Last month, the Office of Science & Technology Policy (OSTP) released their long-expected memo on “Guidelines for Research Security Programs at Covered Institutions.” This memo is the latest action taken by OSTP to implement the requirements in National Security Presidential Memorandum 33 (NSPM-33) and certain provisions of the Chips and Science Act. The purpose is also to, “make sure that institutions of higher education and other research institutions recognize the altered global landscape and fulfill their responsibilities as the first line of defense against improper or illicit activity,” from nation-states and actors.
The memorandum defines a “covered institution” as an organization that is both, “both an institution of higher education, FFRDC, or a nonprofit research institution,” and receives in excess of $50 million per year from the federal government. The memo is then broken into two parts, which correspond to the requirements of covered institutions and the standard requirements of their research security programs, and federal research agencies’ responsibilities and principles for implementation.
In the first part for covered institutions, there are four standard requirements for an institution’s research security program to contain:
Cybersecurity – Requires that institutions of higher education institute a cybersecurity program, “constituent with the cybersecurity resource for research institutions,” within one year after NIST publishes the resource. Non-institutions of higher education are required to certify that they will implement a cybersecurity program, “consistent with another relevant cybersecurity resource maintained by NIST or another federal research agency.”
Foreign travel security – Requires periodic training (at least once every six years) on foreign travel security for covered individuals, “engaged in international travel, including sponsored international travel, for organization business, teaching, conference attendance, or research purposes.” Also requires covered institutions to implement a travel reporting program, “for covered individuals participating in R&D awards when a federal research agency has determined that security risks warrant travel reporting in accordance with the terms of an R&D award.”
Research security training – Institutions are required to implement a research security training program, “for all covered individuals to address the unique needs, challenges, and risk profiles of covered individuals and to certify that the institution ensures that each such covered individual completes such training.” There is some flexibility given to institutions here, as it allows them to use NSF’s training modules or certify that covered researchers have completed a program with similar components.
Export control training – Requires covered institutions to certify that they require, “covered individuals who perform R&D involving export-controlled technologies, to complete training on U.S. export control and compliance requirements.” Again, some flexibility is provided here, allowing institutions to use the training offered by the Bureau of Industry and Security of the Department of Commerce, Directorate of Defense Trade Controls at the Department of State, or a training program with similar components.
In the second part of the memo, there are six responsibilities and principles that research agencies are expected to adhere to:
Non-discrimination – Agencies are to ensure that the research security program requirements they impose, “do not result in targeting, stigmatization, or discrimination against individuals on the basis of race, color, ethnicity, religion, sex (including pregnancy, sexual orientation, or gender identity), national origin, age (40 or older), disability, or genetic information (including family medical history).” There is also a stipulation that agencies require covered institutions to certify that they have implemented safeguards, “to protect the rights of researchers, students, and research support staff or otherwise comply with such requirements.”
Flexibility – Agencies are to allow covered institutions, “to structure their research security program to best serve the institution’s particular needs and to leverage existing programs and activities where relevant, provided that the institution implements all required program components.”
Mechanism for certifications – Requires agencies to provide, “a written or electronic attestation to a federal research agency that the covered institution has met relevant research security program requirements.”
Reducing administrative burdens – In developing their research security program requirements, agencies are expected to, “minimize administrative burden on covered institutions and covered individuals.” Additionally, agencies should encourage, “covered institutions to minimize administrative burden on covered individuals.” There is also a specific call out to be mindful of the administrative burden for less resourced institutions, with EPSCoR, HBCU, and MSI institutions specifically cited.
Minimizing impact to smaller institutions – Straight from the memo: “Federal research agencies should avoid disadvantaging non-covered institutions during the award process in order to facilitate broad participation in the federal R&D enterprise.”
Additional requirements for the agency’s mission/community – NSPM-33 permits agencies to develop additional requirements for their specific mission and community. The memo limits agencies to cases where, 1) policies are required by “statute, regulation, or executive order,” 2) more protections are need for R&D that is, “classified information, technologies subject to Export Administration Regulations, or otherwise legally protected matters,” or 3) “other compelling agency-specific reasons” which are consistent with the law or the agency’s mission.
Finally, the memo begins the implementation timeline for these requirements. Agencies have six months to submit their plans to OSTP and OMB for the purposes of updating their policies, “to ensure this guidance is reflected in the Research Security Programs Standard Requirements of each federal research agency.” The updated policies are expected to go into effect six months after OSTP and OMB sign off on the plans. Agencies are then required to make sure covered institutions have adequate time to implement their research security programs. However, institutions must have their programs up and running no more than 18 months after the effective date of the agencies’ plans. Put another way, the community has no more than two and a half years to start up their research security programs, and those programs start impacting researchers directly.
There are several ways of looking at this document. From a positive perspective, the memo itself is quite reasonable and has no real surprises in it. OSTP and the federal research agencies have telegraphed their actions for the past several years on this topic. The policies set out in the memo provides plenty of lead time for the research community to implement the requirements. As an example, NSF has steadily rolled out their research security actions over the last two years, from data analytics practices, to training modules, and their TRUST risk management framework. The other agencies have done the same or will start taking similar actions.
From another perspective, these requirements will start impacting researchers directly, particularly at universities, very soon. Research security programs are here to stay, and the research community should take these requirements seriously. For more context on the complexities of this topic, in February, several research agencies went before the House Science Committee and spoke about the challenges of implementing these policies and the trepidation coming from their individual communities. Little has changed in the intervening months, except that the policies are now on the path to implementation.
[Editor’s Note: This post was written by CRA’s Tisdale Policy Fellow for Summer 2024, Radhika Agrawal.]
In our ongoing coverage of the Fiscal Year 2025 (FY25) federal budget process, we turn to the House Appropriations Committee’s Energy and Water Development bill. The bill outlines the budget allocations for the Department of Energy’s Office of Science (DOE SC) and ARPA-E. Even though the House Committee has recommended only a small increase for DOE SC overall, there is some good news for the computing community: the ASCR program receives a substantial funding increase.
The bill proposes a 1.8 percent increase for the Office of Science over FY24 enacted levels, bringing the agency’s budget to $8.39 billion for FY25 (an increase of $150 million). Within the Office of Science, the Advanced Scientific Computing Research (ASCR) program, which houses the majority of the computing research at DOE, would see an overall increase of 8.8 percent – going from $1.02 billion in FY24 to $1.11 billion for FY25, under the House’s plan. However, these proposals are still less than the recommended increase in the President’s budget request in March.
The Advanced Research Projects Agency – Energy (ARPA-E), which supports research for developing high-risk energy technologies that address critical economic, environmental, and energy security challenges, is funded at $450 million, as requested by the Administration. That amount represents a cut of $10 million, or 2.2 percent, as compared against the program’s FY24 budget. No policy details were given in the committee’s report for ARPA-E’s budget.
FY24 Final
FY25 PBR
FY25 House
$ Change
% Change
DOE SC Total
$8.24B
$8.60B
$8.39B
+$150M
+1.8%
ASCR
$1.02B
$1.15B
$1.11B
+$90M
+8.8%
ARPA-E
$460M
$450M
$450M
-$10M
-2.2%
In the committee’s report, DOE’s Established Program to Stimulate Competitive Research (EPSCoR) program gets special attention and the appropriators recommend at least $35 million for the program. Additionally, the committee provides not less than $245 million for quantum information sciences, with a breakdown of $120 million for research and $125 million for the five National Quantum Information Science Research Centers. The report also provides up to $15 million for research in support of the Quantum User Expansion for Science and Technology program (QUEST) to facilitate researcher access to the nation’s quantum computing resources. The recommendation provides $10 million for furthering research and development in the field of quantum entanglement networking, including activities to secure communications between energy systems and protect electric grid Supervisory Control and Data Acquisition (SCADA). Within ASCR, the House allocates $330 million to support research in the mathematical, computational, and computer sciences. The committee provides $35 million to support research to develop a new path to energy efficient computing.
While the House appropriators are relatively supportive of research, they are less supportive for workforce development and broadening participation efforts. In the bill’s report, the committee voices its support for the efforts of the Workforce Development for Teachers and Scientist program. This is a program within the Office of Science which ensures, “a sustained pipeline of science, technology, engineering, and mathematics (STEM) workers to meet national goals and objectives.” However, the committee’s allocated budget ($32 million) falls short of the FY24 enacted levels of the program by $8 million. Moreover, the House appropriators provide no funding for the RENEW and FAIR programs, with no further policy justifications.
The House Appropriations Committee approved their bill on July 9. It will now proceed to the full House chamber for passage, where it is likely to be passed. Once the bill clears the House floor, we will have to wait and see how the Senate handles their funding plan for the energy research accounts. As with the other FY25 appropriations bills, we are not expecting the budget to be passed into law until after the upcoming November elections. Please keep checking back for the latest updates.
Do you have expertise in technological issues? Are you interested in how legislation impacts tech issues? There’s a Congressional fellowship for you! Tech Congress, an organization that, “gives talented technologists the opportunity to gain first-hand experience in federal policymaking and shape the future of tech policy,” is accepting applications for its 2025 class of Congressional Innovation Fellows.
The Congressional Innovation Fellowship is a program to, “bridge the divide between Congress and the technology sector.” Tech Congress provides benefits and assistance in placement in a Congressional personal or committee office. The program is looking for both early and mid-career technologists to get hands on experience working in Congress and learn about the policymaking process. This program has been running for several years, having been founded in 2015, and has sent over 119 technologists to Congress.
Tech Congress’ website has more details on what they’re looking for in applicants. If you’re interested in the intersection of public policy and technology, this is a great opportunity to pursue. They have two information calls scheduled for July 16 and July 29. The deadline to submit applications is August 5.
Please use the Category and Archive Filters below, to find older posts. Or you may also use the search bar.
White House Releases National Security Memorandum on Artificial Intelligence
/In: Artificial Intelligence, Defense R&D Highlights, Policy, R&D in the Press /by Brian MosleyToday, the Biden Administration released a National Security Memorandum (NSM) on artificial intelligence. At a high policy level, the NSM directs the executive branch to take steps to, “(1) ensure that the United States leads the world’s development of safe, secure, and trustworthy AI; (2) harness cutting-edge AI technologies to advance the U.S. Government’s national security mission; and (3) advance international consensus and governance around AI.” According to the White House, the NSM’s, “fundamental premise is that advances at the frontier of AI will have significant implications for national security and foreign policy in the near future.”
This memo has been long expected, as it was mentioned at the end of the AI Executive Order, which was released at the end of October 2023.
The NSM takes several steps to advance the use and adoption of AI for national security purposes. From directing improvements to the security and diversity of semiconductor chips supply chain; to “formally designating the AI Safety Institute” at NIST as the primary point of between US industry and the federal government; to doubling down on the NAIRR pilot; to directing the creation of a “Framework to Advance AI Governance and Risk Management in National Security;” to directing the US Government to collaborate with allies and partner nations to, “establish a stable, responsible, and rights-respecting governance framework,” to use AI in ways that adhere to international law and respect human rights. More details can be read in the Fact Sheet released by the White House.
This is a good step to continue the adoption of AI in the operations of the federal government, as well as to do so in a socially responsible way. However, more remains to be done. CRA will monitor the actions taken around AI at the federal level and we will continue to advocate that fundamental AI research, as well as research into its use and applications, must be a key part of any national strategy around this technology.
FY25 Appropriations Update: With End of Fiscal Year Approaching, Congress Rushes to Pass a Stopgap
/In: Funding, FY25 Appropriations, Research /by Brian Mosley
Update 9/26/24: The Senate approved the CR last night and sent the bill to President Biden to be signed into law.
Original Post: With less than a week till the beginning of Fiscal Year 2025 (October 1st), Congress is rushing to pass a continuing resolution (or CR) to avoid a government shut. Over the weekend, Congressional leaders announced a bicameral, bipartisan compromise to pass a relatively clean CR that would extend funding authority, at FY24 levels, to December 20th. This is likely the best-case scenario given the situation, as House Speaker Johnson (R-LA) originally floated a March 2025 CR with a politically toxic voting amendment attached to it. A clean CR, which does not contain politically contentious policy provisions or funding changes, avoids a messy political fight and the possibility of a government shut down until after the election.
However, confidence that Congress will finalize FY25 by December 20th would be misplaced. As we have said repeatedly during the budget process, the outcome of the November elections will directly impact how FY25 is closed out. If the political calculus is relatively unchanged after the election (meaning a split in control of Congress and Democrats retain the White House), it is likely that this Congress will want to close out FY25 to give the 119th Congress a clean slate. However, if either side gains a clear advantage in the election, the likely outcome will be punting the matter into 2025, where that side will have more say over the funding details. We will have to wait for events to play out.
House Science Committee Passes Nine Bills to Support the Advancement of Artificial Intelligence
/In: Artificial Intelligence, Diversity in Computing, Policy, Research, STEM /by Brian MosleyOn September 11th, the House Science, Space, and Technology Committee passed nine pieces of legislation covering a diverse set of topics around artificial intelligence (AI). The bills are now sent to the full House of Representatives for consideration and potentially passage into law.
The nine bills are:
H.R. 5077, the CREATE AI Act
This is perhaps the piece of legislation of most importance to the computing research community, as it would establish the National Artificial Intelligence Research Resource (NAIRR), provide a governing structure for the resource, and authorize it to receive funding. If you’re a regular reader of the Policy Blog, you’ll recognize this bill from previous posts on AI legislation and it was originally introduced in 2023.
NAIRR, a cyberinfrastructure resources proposed by a Congressionally established task force of the same name, was started up as a pilot program by NSF at the beginning of 2024. The goal of NAIRR, which will be run by NSF and overseen by an interagency steering committee, is to provide, “free or low-cost access to datasets and computing resources for development of AI workflows,” helping to democratize the development and use of artificial intelligence. The bill version would authorize NAIRR to receive $430 million a year for five years (FY2025 to FY2030).
Of all the bills the Science Committee moved on the 11th, this is the most likely to become law. That is because there is a Senate version that was passed by the Senate Commerce Committee over the summer and it enjoys wide, bipartisan support in both chambers of Congress. However, expectations on that happening should be kept in check, as there are few legislative working days left in the year and Congress does not have many legislative vehicles to move this to passage. If it were to move, it would likely do so as an amendment in the annual defense policy bill. There are also rumors of an AI supplemental funding bill during the lame duck session of Congress; however, this author would caution that until we know the outcome of the November elections, such an effort is unlikely.
H.R. 9402, the NSF AI Education Act of 2024
This bill support NSF’s education and professional develop mission related to AI. First, it allows NSF to support scholarships and fellowships in AI, and specifically includes community colleges for support. It also directs NSF to support professional development for students, teachers, faculty, and industry professionals. This includes supplements to students and faculty to attain skills, training, or education in AI, as well as fellowships for industry and school professionals.
Next, this legislation establishes a “Centers of AI Excellence” program at NSF. The agency is directed to run the program in coordination with NIST’s Regional Technology Hubs, while also leveraging the NSF Engines program and other NSF efforts it deems necessary and useful. The establishment of such a center is to, “enhance educational outcomes and drive workforce development by integrating artificial intelligence into teaching, learning, and community engagement.” The bill’s language specifically includes community colleges and “area center and technical educational schools” to the list of groups who can take part in the Center program.
The legislation also directs NSF to make awards to promote research regarding teaching models, and materials for AI and its integration into classrooms, teaching, and learning for Pre-K through grade 12 students who are from low-income, rural, or Tribal populations. Finally, the legislation incorporates AI skills development into the National STEM Teacher Corp.
H.R. 9211, the LIFT AI Act
The Lift AI Act would allow NSF to make awards to institutions of higher education or nonprofit organizations to support research activities to develop educational curricula and evaluation methods for AI literacy at the K-12 level. It also allows NSF to carry out these activities through new or existing funding; it does not provide additional funding authorizations.
There is also a sense of Congress clause that talks about the importance of AI literacy in K-12 for the nation’s future workforce and how it underpins the country’s economic and national security. Such clauses don’t have much legal strength, but they provide a useful rhetorical point for the research agencies to say “Congress deems this subject important and that’s why we’re funding these activities.”
H.R. 9403, the Expanding AI Voices Act
This bill directs NSF to broaden participation and capacity in AI research, education, and workforce development. It does this by competitive awards to institutions that are not one of the top 100 institutions (as determined by Federal R&D expenditures during the 3-year period prior to the year of the award), HBCUs, MSIs, tribal colleges or universities, or a consortia of any of the these entities.
An eligible institution may use the funds to carry out:
NSF is then directed, when performing outreach to the community about this program, to take into account all regions of the country, and to especially consider people from underserved communities and groups historically underrepresented in STEM. No additional funding authorization is provided.
H.R. 9215, the Workforce for AI Trust Act
This bill is meant to, “facilitate a workforce of trained experts to build trustworthy AI systems.” The legislative language is split between sections for NSF and NIST.
For NSF, it allows the agency to support graduate and postdoc research fellowships across disciplines. The language includes humanities and social sciences for fields that should be included. The fellowships are meant for the, “integration of ethical and responsible practices and principles into the design, development, training, deployment, evaluation, and understanding of artificial intelligence systems.” The bill also directs NSF to make awards for the development and hosting of intra and inter-institutional workshops on integrating perspectives and skills from multiple disciplines toward the deployment, evaluation, and understanding of AI systems. Finally, it directs NSF to integrate these perspectives into the agency’s peer review process.
For the NIST section of the bill, it amends NIST’s AI mission to have the agency support education and workforce development activities to expand the AI workforce. This includes careers related to helping organizations govern, map, measure, and manage AI related risks, including testing, evaluation, verification, and validation of AI systems.
H.R. 9466, the AI Development Practices Act
This bill directs NIST to “catalog and evaluate emerging practices and norms for communicating certain characteristics of artificial intelligence systems, including relating to transparency, robustness, resilience, security, safety, and usability, and for other purposes.” This is perhaps the most esoteric bill that the committee considered, but it covers an essential function of the federal agency whose mission is to assemble standards for industry. The language makes clear that any guidance that NIST develops must remain voluntary.
H.R. 9497, the AI Advancement and Reliability Act
This bill establishes a “Center for AI Advancement and Reliability” at NIST in order to ensure US leadership in, “research, development, and evaluation of the reliability, robustness, resilience, security, and safety of artificial intelligence systems.” The center is to coordinate with NSF, OSTP, DOE, DOD, DHS, and other departmental secretaries or agencies as considers appropriate. The bill also directs NIST to establish a consortium of stakeholders from academic, industry and civil society. The center is authorized to receive $10 million for Fiscal Year 2025.
H.R. 9197, the Small Business Artificial Intelligence Advancement Act
This is a small business assistance bill that directs NIST to provide guidance to help such entities utilize advances in the AI marketplace.
H.R. 9194, the Nucleic Acid Screening for Biosecurity Act
Finally, the shortest bill the Science Committee considered, it’s a biology sciences bill that touches on the potential impact AI could have on nucleic acid screening sciences.
Six Leading Computing Organizations Call on Congress to Fully Fund the CHIPS & Science Act
/In: CRA, Funding, R&D in the Press, Research, Statements /by Brian MosleyToday six leading organizations — AAAI, ACM, CRA, IEEE-USA, SIAM, and USENIX — representing more than 400,000 people in computing, information technology, science, and innovation across US industry, academia, and government, joined together to call on Congressional leaders of both parties to fully fund the research agencies contained in the CHIPS and Science Act of 2022:
Download the letter.
FY25 Appropriations Update: Senate Appropriators Call for Cuts and Flat Funding for the Defense Research Accounts
/In: Defense R&D Highlights, Funding, FY25 Appropriations, Research /by Brian MosleyAt the beginning of August, the Senate Appropriations Committee completed its work on its slate of Fiscal Year 2025 funding legislation. In their last bundle of bills was the chamber’s version of the Defense appropriations legislation, which handles funding for the Department of Defense. Unfortunately, the Senate’s plans are generally bad for the defense research accounts, calling for cuts or flat funding for basic, applied, and advanced technology development research areas.
As a refresher, DOD’s Science and Technology program is made up of three accounts: 6.1 (basic research), 6.2 (applied research), and 6.3 (advanced technology development). These accounts are themselves made up of individual accounts for each of the three services (Army, Navy, and Air Force), as well as a Defense Wide account. DARPA, or the Defense Advanced Research Projects Agency, is a section under the Defense Wide account.
Getting into the details of the Senate’s proposal, Basic Research (6.1), which is the main Defense Department supporter of fundamental research at US universities, would receive a cut of 4.2 percent compared to its FY24 levels. The account goes from $2.63 billion in FY24 to $2.52 billion for FY25, a decrease of $110 million. A good amount of this decrease is due to reductions in the different services’ “University Research Initiative” subaccounts: the Army’s subaccount decreases by 8.8 percent, the Navy’s by -6.7 percent, and the Air Force’s by -20 percent. Only the Space Force received an increase (2.3 percent). The only other bright spot is the Defense-Wide account, which received an overall increase of about 2.0 percent.
The Applied Research (6.2) account fairs the worst among the defense accounts, receiving a significant cut in funding under the Senate’s mark. The full account would see a 14.1 percent decrease compared to last year’s budget, going from $7.60 billion in FY24 to $6.53 billion under the Senate’s plan, a reduction of $1.07 billion.
The Advanced Technology Development (6.3) account would be flat funded, receiving a slight cut under the Senate’s budget framework. It would go from $11.29 billion in FY24 to $11.20 billion in FY24, a decrease of $90 million (or -0.8 percent).
Finally, DARPA would be the one bright spot for defense research in the Senate’s funding mark. The agency would receive a 17 percent increase and go from $4.12 billion in FY24 to $4.84 billion in FY25, an increase of $720 million.
This is yet another example of a poor Presidential request creating problems further down the budgetary path. When one compares the Senate numbers against the President’s request from March, the 6.1, 6.2, and 6.3 accounts would receive a 2.9, 12.7, and 25 percent increase respectively. Put another way, the Senate looked after these accounts by giving them more than DOD requested, even though they are less than the accounts received a year before. While the Defense Department is in a difficult position with its overall budget, saying back in March that their planning comes from a zero-growth mindset, this is still poor support for the nation’s defense research efforts.
It’s hard to say where a final defense research budget will ultimately settle. The House and Senate are not far apart with their budget plans, which could make final numbers easy to agree. Unfortunately, with both chambers calling for cuts or flat funding, the likely final numbers could be poor. However, it is clear that nothing will be decided by the beginning of the new Fiscal year (October 1st). And as we have said about all the other research agencies, a final FY25 budget won’t be decided until after the November elections. In fact, depending on the outcome of the election, a final resolution for FY25 may slip into the next calendar year. CRA will continue to monitor the situation and report on any new develops; please keep checking back for the latest news.
FY25 Appropriations Update: Senate Appropriators Support the DOE’s Requested Budget; Recommend Increases in AI, QIS, and Other Computing Research Areas
/In: Funding, FY25 Appropriations, Research /by Brian MosleyContinuing our coverage of the Fiscal Year 2025 (FY25) federal budget process, we turn to the Senate Appropriations Committee’s Energy and Water bill. This bill contains the budgets for the Department of Energy’s Office of Science (DOE SC) and ARPA-E. The Senate Appropriators supported the President’s requested budget from March and provided generous funding increases for the Office of Science and ASCR. This is contrast to the lower funding levels the House appropriators recommended with their budget plan.
The Senate bill proposes a 4.7 percent increase for the Office of Science over FY24 enacted levels, bringing the agency’s budget to $8.60 billion for FY25 (an increase of $360 million). Within the Office of Science, the Advanced Scientific Computing Research (ASCR) program, which houses the majority of the computing research at DOE, would see a substantial increase of 12.7 percent – going from $1.02 billion in FY24 to $1.15 billion for FY25. These numbers are substantially the same as what the Department requested for their budgets. The Senate appropriators made a point of calling out scientific discovery and Chips & Science Act implementation as priorities in their bill, saying it, “continues to advance the highest priorities in materials research, high-performance computing, artificial intelligence, biology, and clean energy research to maintain and strengthen our global competitiveness.”
In the committee’s report, the Senate appropriators continued their support for the agency’s efforts in artificial intelligence and quantum information sciences. For AI specially, they are recommending a new line item of $100 million for artificial intelligence and machine learning efforts, and voicing support for the Department’s new Frontiers in Artificial Intelligence for Science, Security, and Technology [FASST] initiative. The appropriators also recommend no less than $160 million for AI efforts across the Office of Science, with a leadership role for ASCR, and encouraged DOE, “to apply those capabilities to the Office of Science’s mission with a focus on accelerating scientific discovery.” Within ASCR, support is voiced for the program’s Mathematical, Computational and Computer Sciences research efforts. Finally, the Senate directs DOE to establish, “a hybrid High-Performance Computing [HPC]/Quantum Computing Pathfinder Program at a Leadership Computing Facility,” in order to allow the LCF to, “to acquire an on-premise quantum computer…to begin to study how to effectively interface and integrate quantum processing units with traditional HPC resources.”
The Senate appropriators also voiced their support for the Office of Science’s engagement plans with HBCUs and MSIs, with particular praise for the agency’s RENEW and FAIR programs. This is in contrast to the House’s plans which eliminated funding for the programs. The Senate budget mark also support’s DOE’s EPSCoR program, recommending $35 million.
Finally, the Advanced Research Projects Agency – Energy, or ARPA-E, would receive a light cut of a million dollar, or a reduction of 0.2 percent compared to FY24 enacted levels. The agency would go from $460 million in FY24 to $459 million under the Senate’s FY25 plans. That is above both the President’s request and the House’s mark.
The Senate Appropriations Committee unanimously approved their Energy and Water legislation on August 1st; the bill now heads to the full chamber for consideration. As we have mentioned before, we are not expecting the Fiscal Year 2025 budget to be completed until after the November election. And the results of that election could mean that finalizing FY25 is punted until the new Congress in the next calendar year. We will have to wait and see what happens. CRA will continue to monitor the situation and report on any new develops; please keep checking back for the latest news.
FY25 Appropriations Update: Senate Marks for NSF, NIST, NASA Released; Contains Increase for NSF, other Agencies, and Expands Funds for Chips Act Implementation
/In: Funding, FY25 Appropriations, Information Technology R&D Highlights, Research /by Brian MosleyIn our ongoing series looking at the Fiscal Year 2025 budget, we turn to the Senate Appropriations Committee and their Commerce, Justice, Science legislation. This bill is of most concern to the computing research community, as it contains the budgets for the National Science Foundation (NSF), the National Institute of Standards and Technology (NIST), and NASA. The Senate numbers are generally better than what the House appropriators recommended in July, while not as good as what the Biden Administration requested in their funding request. This was accomplished by expanding the allocation of funds that are available for this bill, a departure from the House approach, and prioritizing funding for implementing the Chips and Science Act activities around AI and semiconductor development.
Getting into the numbers, under the Senate’s plan NSF would receive a 5.4 percent increase, going from $9.06 billion in FY24, to $9.55 billion in FY25, an increase of $490 million. While this is a good number, it is still below the levels set in the Fiscal Year 2023 budget, and what the Biden Administration requested in March.
The increase to NSF is evenly distributed throughout the agency. The Research and Related Activities (R&RA) account, which houses the agency’s research portfolio, would receive $7.53 billion for FY25, which is an increase of 4.9 percent or a plus up of $350 million. The Directorate for STEM Education (EDU) would receive $1.23 billion, which is an increase of $60 million from FY24, or +5.1 percent. These numbers are generally better than what the House appropriators recommend under their funding legislation, with the exception of RRA, which the House prioritized and provided slightly more funds.
The policy provisions contained in the committee’s report had several topics related to computing issues in the section devoted to RRA. Artificial intelligence features prominently, and the Senate appropriators provide “not less than” the FY24 funding levels for AI research. Additionally, the committee, “encourages NSF to fund meritorious research and develop technical methods and techniques to improve the transparency, interpretability, and explainability of AI to better understand why and how models arrive at their decisions, recommendations, and other outputs.” Also, the National AI Research Resource (NAIRR) was provided with $30 million in funding. With regard to quantum research, the committee fully funds the President’s requested budget and, “encourages NSF to partner with institutions of higher education, industry, and other Federal agencies in order to develop the next generation of quantum computing workforce,” using grants and interdisciplinary research initiatives geared toward workforce development. The committee continues to support the TIP Directorate and reminded NSF that the success of the directorate, “will be enhanced through investing in the necessary foundational basic research provided by scientific disciplines across the research spectrum.” Within that, the Senate provides $200 million for the NSF Engines program and directs the agency to award at least 20 percent of the program’s funds in EPSCoR states. In a related area, the Senate appropriators directed NSF, “to the maximum extent practicable, 16.5 percent of NSF research funding and 20 percent of scholarship funding” should go to EPSCoR States; this is in line with the policy set out in the Chips and Science Act. Finally, the committee voiced its support for NSF’s research security efforts and encourages the agency to continue to work with the community on implementation.
With regard to provisions for the STEM Education Directorate, the Senate appropriators are setting up a clash with their House counterparts on several topics. For example, the Senate Committee “maintains its strong support for NSF’s informal science education program,” and provides $70 million; the House appropriators zeroed out the budget for this program under their plan. The Senate appropriators also make a point of voicing its support for NSF’s broadening participation efforts.
The National Institute of Standards and Technology (NIST) would continue to have a complicated budget under the Senate’s FY25 funding plan. At the topline NIST would receive $1.53 billion for FY25, a 4.8 percent increase, or $70 million more than it received in FY24 ($1.46 billion). However, the Science and Technical Research and Services (STRS), would receive $1.06 billion, a decrease of $20 million, or -1.9 percent, compared to FY24. STRS contains the majority of the agency’s research portfolio. Both of those numbers are better than what the Administration requested in March and the House approved in July.
Regular readers will recall that NIST’s budget situation has become complicated over the last several years due to the large number of earmarks Congress has approved in its budget. The Senate’s proposed funding plan is no different. This makes any apples-to-apples comparison of NIST’s budget difficult. When the earmarks are removed, STRS would receive an increase of over 13 percent compared to FY24 levels (these numbers are tracked by AIP FYI).
In terms of policy provisions, the committee provides up to the requested levels to, “advance AI research, standards, testing, and the U.S. AI Safety Institute [USAISI],” at the agency. The Senate appropriators were also highly complementary of the creation of the USAISI, and that NIST sought public input from the community on several topics and directs the agency to continue these efforts. With regard to quantum information sciences research, the committee provides not less than the FY24 levels for these efforts. And the committee’s report contains several other computing related topics such as cybersecurity, digital twin technology, and firmware resiliency.
Finally, NASA receives a slight increase under the Senate’s plans. At the topline, NASA would receive $25.4 billion for FY25, an increase of $550 million from FY24 levels (+2.0 percent). The agency’s Science account would receive a slight better increase than the topline, receiving $7.57 billion, an increase of 3.3 percent, or +$240 million, over FY24 ($7.33 billion).
The Senate Appropriations Committee passed the CJS bill with bipartisan support on July 25th; it is unclear when or if it will be considered by the full Senate chamber. With the House CJS legislation passed in July, technically negotiations to finalize these budgets can begin between the two chambers. However, no one in Washington believes that FY25 will be completed until after the November general election. And the outcome of that election will heavily influence how things are handled; if one party or the other comes out with a clear advantage, final consideration of the budget will likely to be punted into the 2025 calendar year. We still have a long way until Fiscal Year 2025 is completed. CRA is continuing to monitor the situation for any developments, so please check back for updates.
OSTP Releases Research Security Memo to Research Agencies; Begins Implementation Timeline
/In: Impediments to Research Highlights, Information Technology R&D Highlights, Policy, Research /by Brian MosleyLast month, the Office of Science & Technology Policy (OSTP) released their long-expected memo on “Guidelines for Research Security Programs at Covered Institutions.” This memo is the latest action taken by OSTP to implement the requirements in National Security Presidential Memorandum 33 (NSPM-33) and certain provisions of the Chips and Science Act. The purpose is also to, “make sure that institutions of higher education and other research institutions recognize the altered global landscape and fulfill their responsibilities as the first line of defense against improper or illicit activity,” from nation-states and actors.
The memorandum defines a “covered institution” as an organization that is both, “both an institution of higher education, FFRDC, or a nonprofit research institution,” and receives in excess of $50 million per year from the federal government. The memo is then broken into two parts, which correspond to the requirements of covered institutions and the standard requirements of their research security programs, and federal research agencies’ responsibilities and principles for implementation.
In the first part for covered institutions, there are four standard requirements for an institution’s research security program to contain:
In the second part of the memo, there are six responsibilities and principles that research agencies are expected to adhere to:
Finally, the memo begins the implementation timeline for these requirements. Agencies have six months to submit their plans to OSTP and OMB for the purposes of updating their policies, “to ensure this guidance is reflected in the Research Security Programs Standard Requirements of each federal research agency.” The updated policies are expected to go into effect six months after OSTP and OMB sign off on the plans. Agencies are then required to make sure covered institutions have adequate time to implement their research security programs. However, institutions must have their programs up and running no more than 18 months after the effective date of the agencies’ plans. Put another way, the community has no more than two and a half years to start up their research security programs, and those programs start impacting researchers directly.
There are several ways of looking at this document. From a positive perspective, the memo itself is quite reasonable and has no real surprises in it. OSTP and the federal research agencies have telegraphed their actions for the past several years on this topic. The policies set out in the memo provides plenty of lead time for the research community to implement the requirements. As an example, NSF has steadily rolled out their research security actions over the last two years, from data analytics practices, to training modules, and their TRUST risk management framework. The other agencies have done the same or will start taking similar actions.
From another perspective, these requirements will start impacting researchers directly, particularly at universities, very soon. Research security programs are here to stay, and the research community should take these requirements seriously. For more context on the complexities of this topic, in February, several research agencies went before the House Science Committee and spoke about the challenges of implementing these policies and the trepidation coming from their individual communities. Little has changed in the intervening months, except that the policies are now on the path to implementation.
FY25 Appropriations Update: House Appropriators Recommend Flat Funding the Office of Science but Give a Healthy Increase to ASCR
/In: Funding, FY25 Appropriations, Information Technology R&D Highlights, Research /by Brian Mosley[Editor’s Note: This post was written by CRA’s Tisdale Policy Fellow for Summer 2024, Radhika Agrawal.]
In our ongoing coverage of the Fiscal Year 2025 (FY25) federal budget process, we turn to the House Appropriations Committee’s Energy and Water Development bill. The bill outlines the budget allocations for the Department of Energy’s Office of Science (DOE SC) and ARPA-E. Even though the House Committee has recommended only a small increase for DOE SC overall, there is some good news for the computing community: the ASCR program receives a substantial funding increase.
The bill proposes a 1.8 percent increase for the Office of Science over FY24 enacted levels, bringing the agency’s budget to $8.39 billion for FY25 (an increase of $150 million). Within the Office of Science, the Advanced Scientific Computing Research (ASCR) program, which houses the majority of the computing research at DOE, would see an overall increase of 8.8 percent – going from $1.02 billion in FY24 to $1.11 billion for FY25, under the House’s plan. However, these proposals are still less than the recommended increase in the President’s budget request in March.
The Advanced Research Projects Agency – Energy (ARPA-E), which supports research for developing high-risk energy technologies that address critical economic, environmental, and energy security challenges, is funded at $450 million, as requested by the Administration. That amount represents a cut of $10 million, or 2.2 percent, as compared against the program’s FY24 budget. No policy details were given in the committee’s report for ARPA-E’s budget.
In the committee’s report, DOE’s Established Program to Stimulate Competitive Research (EPSCoR) program gets special attention and the appropriators recommend at least $35 million for the program. Additionally, the committee provides not less than $245 million for quantum information sciences, with a breakdown of $120 million for research and $125 million for the five National Quantum Information Science Research Centers. The report also provides up to $15 million for research in support of the Quantum User Expansion for Science and Technology program (QUEST) to facilitate researcher access to the nation’s quantum computing resources. The recommendation provides $10 million for furthering research and development in the field of quantum entanglement networking, including activities to secure communications between energy systems and protect electric grid Supervisory Control and Data Acquisition (SCADA). Within ASCR, the House allocates $330 million to support research in the mathematical, computational, and computer sciences. The committee provides $35 million to support research to develop a new path to energy efficient computing.
While the House appropriators are relatively supportive of research, they are less supportive for workforce development and broadening participation efforts. In the bill’s report, the committee voices its support for the efforts of the Workforce Development for Teachers and Scientist program. This is a program within the Office of Science which ensures, “a sustained pipeline of science, technology, engineering, and mathematics (STEM) workers to meet national goals and objectives.” However, the committee’s allocated budget ($32 million) falls short of the FY24 enacted levels of the program by $8 million. Moreover, the House appropriators provide no funding for the RENEW and FAIR programs, with no further policy justifications.
The House Appropriations Committee approved their bill on July 9. It will now proceed to the full House chamber for passage, where it is likely to be passed. Once the bill clears the House floor, we will have to wait and see how the Senate handles their funding plan for the energy research accounts. As with the other FY25 appropriations bills, we are not expecting the budget to be passed into law until after the upcoming November elections. Please keep checking back for the latest updates.
Congressional Innovation Fellowships Accepting Applications for 2025 Class; Deadline to Apply August 5
/In: General, Misc., People /by Brian MosleyDo you have expertise in technological issues? Are you interested in how legislation impacts tech issues? There’s a Congressional fellowship for you! Tech Congress, an organization that, “gives talented technologists the opportunity to gain first-hand experience in federal policymaking and shape the future of tech policy,” is accepting applications for its 2025 class of Congressional Innovation Fellows.
The Congressional Innovation Fellowship is a program to, “bridge the divide between Congress and the technology sector.” Tech Congress provides benefits and assistance in placement in a Congressional personal or committee office. The program is looking for both early and mid-career technologists to get hands on experience working in Congress and learn about the policymaking process. This program has been running for several years, having been founded in 2015, and has sent over 119 technologists to Congress.
Tech Congress’ website has more details on what they’re looking for in applicants. If you’re interested in the intersection of public policy and technology, this is a great opportunity to pursue. They have two information calls scheduled for July 16 and July 29. The deadline to submit applications is August 5.