In a recent speech, the director of the White House Office of Science and Technology Policy (OSTP) declared that promoting America’s technological leadership requires three things of government: (1) to creatively allocate public research and development dollars, (2) to construct a common-sense, pro-innovation regulatory regime, and (3) to adopt the incredible products and tools made by American builders—all in support of shaping a funding environment tied to national priorities. Additionally, a recent Executive Order targeting the Federal Acquisition Regulation (FAR) seeks to revamp the federal procurement system into one that is more prudent, agile, and efficient. While the Trump Administration promises an unbridled innovation economy for promoting progress and sustaining American leadership in AI, recent proposed cuts to federal funding under the auspices of government efficiency will hamstring the broader research and development (R&D) ecosystem at a time when collaboration between the academic, private, and public sectors is critically needed. However, there exists at least one policy mechanism that incentivizes and encourages AI innovation and satisfies the OSTP director’s three requirements: federal prize competitions.
Federal prize competitions are contests sponsored or facilitated by one or more government agencies, sometimes in partnership with industry or other external entities, in which the government uses monetary and nonmonetary incentives to either (a) advance knowledge or benchmarks within a particular field or (b) solicit tools or solutions for specific problems. The benefits of competitions differ from traditional R&D, funding schemas, or contract awards. First, competitions shift the risk of failure to participants and usually pay only for successes, which means the sponsoring agency can establish ambitious goals. Second, competitions provide government agencies with a means for testing the operational effectiveness of tools, solutions, or approaches before any potential prototyping, scaling, or procurement decisions. In some cases, the competition itself may also serve as a vehicle for faster acquisition. Third, competitions typically have relatively low barriers to entry, which allows for a broader and more diverse participant pool.
In 2020, U.S. departments and agencies sponsored 17 AI- and cyber-specific federal prize competitions valued at $5.8 million. Perhaps unsurprisingly, the majority of these challenges were sponsored by the Department of Defense (DOD). In 2024, U.S. departments and agencies sponsored 27 AI- and cyber-specific federal prize competitions with both monetary and non-monetary incentives totaling over $36 million.1 This represents a nearly 60% increase in activity and a 520% increase in monetary awards over only four years. As of 2024, the DOD still led in sponsorship. The Department of Health and Human Services (HHS) and the Department of Energy (DOE) also historically rank among the top federal departments and agencies that sponsor prize challenges.
Table 1. 2024 AI and Cyber Federal Prize Competitions
| Department or Agency | Number of Challenges | Total Monetary Award |
|---|---|---|
| Department of Defense | 19 | $32,843,000 |
| Department of Health and Human Services | 2 | $50,000 |
| Department of Energy | 1 | $1,850,000 |
| Federal Trade Commission | 1 | $35,000 |
| Department of Veterans Affairs | 1 | $1,000,000 |
| National Institute for Standards and Technology | 1 | $17,500 |
| Small Business Administration | 1 | $350,000 |
| National Aeronautics and Space Administration | 1 | No Monetary Prize Awarded |
| Total | 27 | $36,145,500 |
To truly empower American innovation in AI, federal departments and agencies need to use all tools at their disposal. Aside from the benefits outlined above, there are three main reasons why continued support of federal prize competitions would augment the Trump administration’s innovation, R&D, and procurement priorities.
Competitions advance innovation at a fraction of the typical costs. Historically, federal prize competitions have generated novel insights and solutions for fractions of typical R&D costs. For example, out of 814 federal prize competitions between 2010 and 2020, the total prize money awarded was roughly $243 million. This is a small cost compared to total federal R&D outlays of $1.3 trillion over the same time period. However, prize competitions can only augment R&D and are not meant to replace it. These are tools best suited for chasing specific, measurable, or solvable challenges. Also, prizes and awards do not need to be monetary and are only paid out when the challenge requirements are met or satisfied. Some competitions have offered award packages that include R&D mentorship, access to federal facilities or laboratories for the purposes of completing the challenge or incubation, or opportunities to pitch solutions in front of larger federal audiences.
On the other hand, monetary awards do not have to have significant value to be effective. In 2020, a General Services Administration (GSA) challenge sought AI/ML enabled solutions to automatically review end-user license agreements for terms and conditions that were unfavorable to the federal government and could shorten the length of time it took for contracting officers to review, negotiate, and accept them. The competition produced three winning solutions that GSA was able to license—with a total prize purse of $20,000.
Competitions have bipartisan support. Federal prize competitions have bipartisan support. In 2024, Sens. Booker (D-NJ), Rounds (R-SD), and Heinrich (D-NM) announced the AI Grand Challenges Act. The legislation directs the National Science Foundation (NSF) to establish a challenge program to administer federal prize competitions to solve or advance AI-specific problems across a range of sectors from energy to healthcare. Additionally, federal prize competitions are supported by the America COMPETES Reauthorization Act, which extends broader authorities to all federal agencies and departments to facilitate prize challenges that further agency-specific missions.
Competitions get results. Prize competitions have resulted in major breakthroughs or advancements, and have led to workable solutions for federal agencies. Most famously, the Defense Advanced Research Projects Agency’s (DARPA) 2016 Cyber Grand Challenge advanced the field of automated vulnerability discovery and patching. The winner, a product called “Mayhem,” was later awarded an $8 million Defense Innovation Unit (DIU) contract to take Mayhem to production and a $45 million DOD contract to deploy Mayhem across DOD networks.
While the Cyber Grand Challenge was one singular competition, others can be iterative. For example, the Army’s xTech Challenge program is a series of various prize challenges that allow businesses and researchers to build, test, and pitch solutions specific to Army and DOD needs. One of the challenge series, xTechScalable AI, prioritizes the development of disruptive AI solutions. The challenge is aligned with Project Linchpin, which is an effort to deliver trusted AI and Machine Learning (ML) capabilities to the Army’s electronic warfare programs. These programs allow the DOD to integrate non-traditional vendors—yet another benefit of prize competitions—into its R&D ecosystem by awarding Small Business Innovation Research (SBIR) Direct to Phase II contracts to qualified contestants. Last year’s winner, Protopia AI, developed privacy enhancements and data protections for generative AI and Large Language Models (LLMs) using their proprietary platform.
Not only does this allow government agencies to test potential solutions, but it also opens doors for collaboration and partnership, especially at a time when partnership between government and the research community is called into question. Last year’s Cyber Innovators Challenge focused on advancing cybersecurity capabilities to address cyber threats faced by the DOD. With a total prize allocation of only $350,000 but the potential for follow-on opportunities like Other Transaction Agreements (OTAs) or Cooperative Research and Development Agreements (CRADAs), the Cyber Innovators Challenge generated promising innovations. For example, one of the winners built Honeypot LLM, which raises the cost of would-be hackers by simulating potential victims using socially engineered, conversational interactions to extract adversaries’ tactics, techniques, and procedures.
It is also worth noting that many competitions are structured to offer follow-on opportunities. Prize competitions can be organized and structured to satisfy competitive procurement requirements and thus streamline the process. This is known as challenge-based acquisition, which MITRE defines as “using challenges to communicate needed capability, encourage innovation in a minimally prescriptive environment, assess candidate offerings, and purchase solutions.”
Using government-sponsored prize challenges to encourage innovation or incentivize progress is not unique to the United States. Most notably, China significantly increased its efforts to facilitate state-run capture-the-flag (CTF) competitions following the DARPA Cyber Grand Challenge in 2016. Since then, China has developed and supported a thriving CTF ecosystem that builds community, bolsters the workforce, stimulates innovation, and serves as a talent resource. This ecosystem has become essential to China’s domestic capability development and aligns with policy directives that are part of President Xi’s vision of becoming a “cyber powerhouse.”
Additionally, both Japan and South Korea are using prize challenges to spur domestic innovation and collaboration. Japan’s Generative AI Accelerator Challenge is aimed at building domestic platform model development capability and to encourage private sector collaboration. While South Korea’s approach is more focused on developing its own version of ChatGPT using significant government investment, there are also plans to convene a global AI challenge initiative to foster domestic talent and attract international researchers.
A pro-growth American innovation economy is simply not possible without all parts working in tandem to sustain American leadership and competitiveness. While prize competitions are not a suitable replacement for basic and applied research, they can be used as vehicles to encourage private sector innovation in areas of national security or economic interest. There are rarely failures. Competitions that fail to produce results or winners can instead be viewed as establishing benchmarks for the field or tool at a relatively modest cost. Furthermore, prize competitions can remain a critical and necessary bridge between the U.S. innovation ecosystem—industry, academia, and the federal government—at a time when tensions and other policy decisions might prevent collaboration. Competitions are gaining momentum, but are still widely underutilized as tools. As such, it is imperative that the Trump administration recognizes the power and potential of the policy mechanisms at its disposal to meet its promise of unleashing American innovation.
- To determine which challenges were specific to AI or Cyber, four CSET researchers examined all 129 federal prize competitions from 2024 and flagged challenges that either (a) explicitly mentioned AI or cyber in the description or (b) explicitly mentioned the development of algorithms, models, or using data to make informed autonomous decisions. Challenges that were vague or open-ended but could potentially include AI or Cyber, such as those calling for ‘innovative solutions,’ were excluded.