ACL Caregiver AI Prize Challenge — Phase 1 (2026)

ACL Caregiver AI Prize Challenge — Phase 1 (2026)

Historical reference: ACL’s Phase 1 application deadline was July 31, 2026, at 5:00 p.m. ET. That submission window has closed. ACL has not published a new application deadline or Phase 2/Phase 3 prize amount, so this page does not present the opportunity as currently open.

The ACL Caregiver AI Prize Challenge was a federal prize competition administered by the Administration for Community Living (ACL), an operating division of the U.S. Department of Health and Human Services. It sought practical artificial-intelligence tools that could help people provide care at home and in community settings. The intended beneficiaries included family caregivers, friends and neighbors who provide support, paid direct care workers, home-care organizations, and people receiving care.

This was not a conventional grant solicitation. Applicants competed against judging criteria, and the Phase 1 awards were prizes for selected designs. The official challenge structure had two tracks running concurrently and three progressive phases. Phase 1 focused on design: an applicant had to show a credible AI-enabled solution, evidence that the underlying technology had reached at least Technology Readiness Level 3, and a realistic path toward implementation and testing. Winners could then move into Phase 2, while only Phase 2 winners could compete in Phase 3.

Key details

What the challenge was designed to support

ACL framed the competition around the strain placed on people who provide care and the shortage of paid caregivers. The agency was interested in tools that make caregiving more sustainable, affordable, safe, and workable in ordinary homes and community environments. Examples on the official page include tools that detect meaningful changes in a person’s condition, help coordinate transportation, support scheduling, improve communication among caregivers and professionals, reduce documentation work, and provide training or on-demand assistance.

The focus was not simply on adding an AI feature to an existing product. A strong submission needed to connect a specific caregiving problem to a specific workflow and explain how the proposed tool would improve that workflow. The challenge materials repeatedly emphasized caregiver input, usability, affordability, privacy, dignity, choice, and human accountability. A system that saves administrative time but makes it harder for a caregiver to understand or correct an output would not fit the program’s principles well.

ACL also required a direct connection to caregiver experience. A product intended only to improve the independence of a person with a disability, without explaining how it supports or changes the caregiver’s work, was not enough for the challenge. Solutions could be tailored to particular populations or care needs, including caregivers supporting people with disabilities, older adults, neurodivergent people, or people with condition-specific needs, as long as the relationship to caregiving was clear.

The two tracks

Track 1: AI tools to support caregivers

Track 1 covered AI-enabled tools used by caregivers while supporting an individual in a home or community setting. The solution could assist family caregivers, friends, neighbors, or the direct care workforce. ACL’s judging framework looked for a clear understanding of the caregiver problem, a viable response to that problem, a plan based on caregiver and care-recipient input, and a design that supports rather than replaces human judgment.

Track 1 applications also needed an implementation approach. The Phase 1 design did not have to be a finished consumer product, but it did need to show how the team would move from a working proof of concept toward testing in a realistic caregiving environment. The plan was expected to address performance measurement, user feedback, adaptation, usability, safety, and integration with relevant systems when applicable.

Track 2: AI tools for extending the caregiver workforce

Track 2 focused on organizations that help operate or support paid caregiving, such as direct care employers, home-care cooperatives, state agencies administering community-based care, and other home- and community-based providers. Relevant uses included reducing documentation burden, improving staff scheduling and deployment, increasing time spent with care recipients, and supporting recruitment, retention, or training.

Track 2 still required the same core principles. Operational efficiency alone was not the whole test. Applicants had to show that the proposed tool would create value for direct care workers and the people they serve, use input from end users, promote high-quality care, and minimize harmful, biased, or unsafe AI behavior.

Phase 1 prize amounts

The official ACL track descriptions stated that up to 10 applications in each track could receive up to $100,000 each. Each track also allowed up to five meritorious awards of up to $50,000 for solutions addressing specified focus areas. Those areas included support for caregivers of people with intellectual and developmental disabilities, support for caregivers of people with Alzheimer’s disease and related dementias, interoperability with electronic medical records or home-device systems, interoperability with assistive technology, and collaborations that go beyond a single-vendor solution. The workforce track included a related collaboration focus for workforce needs.

The HHS launch announcement described Phase 1 as offering up to $2.5 million in prize funding and as many as 20 winners across the two tracks. The USAGov listing likewise described total cash prizes of $2.5 million. The more detailed ACL track pages supplied the per-track award limits and the meritorious-award structure. Prize amounts for Phases 2 and 3 were not announced in the official FAQ and challenge material reviewed for this archive entry.

The prize was not a budgeted grant. ACL’s FAQ stated that applicants did not need to submit a budget and that there were no special requirements for how prize money had to be used. Prize payments could be subject to federal income tax. ACL encouraged, but did not require, participants to obtain a free Unique Entity ID through SAM.gov because it could help with prize payment. For a winning team, the cash prize would be paid to the designated team leader, who would be responsible for distributing funds among team members. An entity winner would be paid directly.

Eligibility rules

The challenge allowed participation by an individual, an entity, or a team, but the award rules were specific. An individual had to be a U.S. citizen or permanent resident. A team leader also had to be a U.S. citizen or permanent resident. Non-U.S. citizens and non-permanent residents could participate as members of an otherwise eligible team or entity, but they could not receive a monetary prize. A private entity had to be incorporated in the United States and maintain its primary place of business in the United States.

Participants had to be at least 18 years old at the time of submission. Federal entities and federal employees acting within the scope of their employment were not eligible in the ordinary way; non-HHS federal employees were directed to consult their agency ethics officials about possible restrictions. HHS employees acting in a personal capacity were excluded by the rules. Judges and people involved in the design, production, execution, or distribution of the challenge, along with specified immediate family members, were also excluded.

Federal funding rules mattered. A federal grantee could not use federal funds to develop a challenge application unless that use was consistent with the purpose of the grant and otherwise permitted. A federal contractor could not use federal contract funds to develop an application or support a submission. Teams also had to comply with applicable laws, accept that judges’ decisions were final and binding, and identify business-confidential information appropriately because submissions could be subject to a Freedom of Information Act request.

Technology readiness was an eligibility issue, not merely a scoring preference. ACL required an AI-enabled tool at Technology Readiness Level 3 or higher. In practical terms, an idea on paper was insufficient. The applicant needed a working model, prototype, or set of controlled tests that supported the technical concept. A complete production application or a fully tested tool in a person’s home was not required for Phase 1, but the applicant needed a credible technical basis and an implementation plan for moving toward real-world testing.

How Phase 1 applications were submitted

The official process had an optional Intent to Apply step and a required Phase 1 application. The intent message was meant to help ACL understand who planned to participate. It asked for the team name, team members, primary contact information, intended track, and a brief description of the proposed AI solution and its possible caregiver impact. Submitting an intent did not replace the application, and the information was described as tentative.

For the required submission, the designated primary point of contact had to email the completed materials to CaregiverAI@acl.hhs.gov by the stated deadline. The primary contact had to be a U.S. citizen or permanent resident who was at least 18 years old and was responsible for challenge correspondence. If a team submitted more than one application, each complete application had to be sent in its own email.

The application package had to be written in English and use the exact section headings in ACL’s application outline. It had to be submitted as a 508-compliant PDF or Microsoft Word document. The cover page could not exceed one page, the project narrative could not exceed 15 pages, and appendices could not exceed 10 pages. Optional Data Output Logs could be supplied separately and did not count toward those page limits. The package could use one-inch margins and single spacing, a widely available font, and at least 11-point type, with a lower minimum allowed for charts and tables. Applicants could include external links, but the narrative and appendices still had to stand on their own and links had to be accessible without a login. Government logos and official seals were not permitted.

Because the deadline has passed, these steps describe the closed Phase 1 process for historical and planning purposes. The official page did not announce a late-submission or rolling-application route. Readers should not email an old Phase 1 package expecting it to be accepted, and they should not treat the archived deadline as a current invitation.

What reviewers were looking for

ACL’s judging criteria organized review around need, impact, caregiver input, co-implementation, deployment readiness, metrics, evaluation, usability, integration, partnerships, and the Caregiver AI principles. A useful application would have tied each design choice to an observed caregiver problem rather than relying on broad claims about AI. It would have explained what information the system receives, what analysis it performs, what action it recommends, and what the caregiver can do when the recommendation is weak, ambiguous, or wrong.

Responsible design was central. The official principles called for protection of privacy, dignity, and choice; clear limits on data collection and access; human-in-the-loop accountability; support for caregiver well-being; and preservation of human connection. Reviewers also considered whether a tool was transparent, usable in realistic conditions, affordable enough to reach its intended users, and capable of reducing user error rather than simply labeling risks after the fact.

Teams should have prepared controlled test evidence for their Technology Readiness Level 3 claim. ACL’s technology-readiness guidance suggested documenting the engineering basis, experimental validation, relevant constraints, critical technology elements, reproducible results, and a traceable record of design data. It also encouraged workflow maps, human-review and override behavior, an explicit response when the system does not know, and raw performance evidence appropriate to the proposed tool. These expectations made a narrow, testable use case more credible than a large list of unvalidated features.

What happens after Phase 1

The challenge was organized as a progression from design to implementation and testing, then to scalability and sustainability. Phase 1 winners were automatically qualified to apply to Phase 2. Applicants could not enter Phase 2 or Phase 3 directly. Phase 2 was intended to collect performance data, refine the selected solutions with user input, and carry out approved testing plans with technical assistance. Phase 3 was intended to examine broader implementation and long-term sustainability, with public recognition for top solutions.

ACL’s official FAQ stated that the prize amounts for Phases 2 and 3 had not been announced and that most details for those phases had not been released. The agency expected additional information before Phase 2 and planned to finalize the Phase 2 timeline after Phase 1. That is why this page keeps the real Phase 1 deadline in the structured metadata, marks the entry as historical, and does not substitute a guessed future deadline or a claim that applications are open.

Official sources

• ACL Caregiver AI Challenge: https://acl.gov/caregiver-ai-challenge
• ACL Caregiver AI Challenge Application Outline: https://acl.gov/caregiver-ai-application-outline
• ACL Caregiver AI Challenge Definitions, FAQs, and Resources: https://acl.gov/caregiver-ai-definitions-faq
• HHS announcement of Phase 1: https://www.hhs.gov/press-room/acl-launches-phase-1-caregiver-ai-prize-competition.html
• USAGov challenge listing: https://www.usa.gov/challenges/acl-caregiver-ai-prize

The ACL challenge page is the controlling source for any future announcement, amendment, winner notice, or new application instructions. Until ACL publishes a new official phase schedule, this entry should be read as an archive of the closed Phase 1 opportunity, not as an active listing.