Grants for AI in Education 2026: Win Part of $400,000 from the Stanford Create+AI Challenge

Grants for AI in Education 2026: Win Part of $400,000 from the Stanford Create+AI Challenge

The Stanford Accelerator for Learning and Google.org’s Create+AI Challenge was a closed 2026 challenge grant for projects that use artificial intelligence to advance learning, augment teaching, and expand opportunity. The program’s central question was how AI could expand human potential rather than simply automate existing education work. Stanford’s current official page presents the challenge as closed and lists the awardees, so this page is a historical reference rather than an open application listing. No next cycle is announced on the current program page.

The final application deadline was January 12, 2026, at noon Pacific Time. The date is retained in the metadata because it is the real closing date for this cycle; it should not be read as a current invitation to apply. The former application form is closed. Teams interested in a future Stanford call should monitor the official Create+AI Challenge page and confirm any new eligibility, dates, and submission instructions there.

At a glance

What this opportunity is trying to fund

This was not a broad “build anything with AI” grant. Stanford described the challenge as a search for projects that put educators and learners at the heart of AI design. The intended benefits included greater access, agency, connection, learning, well-being, and opportunity. That focus made the call narrower than a general technology competition: a proposal needed to explain the human outcome, not only the model or product.

Across all three tracks, the underlying idea is the same: if AI is used, it should help a person perform better, understand better, or participate more meaningfully. Programs that frame AI as a replacement for teachers, counsellors, or mentors are a poorer fit in this design.

The challenge offered $400,000 in total funding across multiple awards. Stanford’s official description specifies two $50,000 awards in each of the three tracks, plus multiple additional awards ranging from $10,000 to $20,000 across all tracks. The amount was therefore a pool spread across selected projects, not a single $400,000 award and not a guaranteed amount for every finalist.

Funding came with non-cash support. The official page lists mentorship and connections with Stanford faculty, researchers, technologists, and collaborators, including learning-science and teacher co-design workshops. It also lists visibility at the AI+Education Summit on February 10 and 11, 2026, and a potential invitation for some projects to continue development at Stanford in a summer 2026 cohort. Those benefits were part of the challenge structure, but the page does not promise that every awardee received every form of support.

Who the 2026 call was for

Stanford’s official program page identifies the eligible audience as educators, researchers, technologists, nonprofits, and students. The official Stanford newsletter describing the application also encouraged designers and entrepreneurs to apply. The call welcomed both early-stage ideas and existing projects that were ready to pilot, scale, or study.

One team member needed a Stanford affiliation: a current student, scholar, staff member, or alum. The newsletter also stated that international applicants were eligible, with select country restrictions. This is the verified cycle-specific eligibility information. The current official page does not state a general applicant age requirement, so the earlier claim that applicants had to be at least 18 has been removed.

The strongest fit would have had most of the following characteristics:

• You have a concrete problem in education, teaching, workforce development, or learner support that AI can augment.
• Your team includes the required Stanford-affiliated member.
• You can define what “success” looks like in measurable terms and are willing to measure it.
• Your project involves real classroom, workforce, or community settings, not only a theoretical concept.
• You are prepared for a review process that rewards clarity, learning science alignment, and fairness over flashy branding.

A poor fit would have been:

• You are building a pure automation product and do not plan to keep a human in the loop.
• You cannot define a measurable learning, participation, or outcomes metric.
• Your team has no member with the Stanford affiliation required for this cycle.
• You are looking for a vague “pay us now, we will figure it out” grant and are not ready to design next-step pilots.

These fit judgments are useful for future calls, but they do not reopen this one.

Readiness check before you spend time

1) Is your use case truly human-augmentation?

Examples that tend to fit:

• AI helps teachers spend less time on repetitive grading admin, while preserving instructional judgment.
• AI gives learners with disabilities accessible ways to participate and express understanding.
• AI helps mentors scale meaningful career guidance with guardrails.

Examples that likely do not fit:

• AI auto-generates final educational content with no teacher oversight.
• A product that mainly replaces existing human support roles.
• A prototype without clear educational benefit beyond curiosity.

2) Can you measure impact in a way that a reviewer can verify?

The scorecard mentions measurement, and Stanford’s listed criteria also emphasize feasibility. A credible plan at minimum should explain:

• what you will measure,
• who you will measure it with,
• and how you will collect the signal ethically.

For example, participation lift, confidence scores, mastery changes, attendance pattern changes, mentor follow-through rates, or pilot retention may be acceptable if definitions and timeframe are realistic.

3) Can you support fairness and accessibility?

This is a serious filter. Even if not listed as a separate checkbox everywhere, fairness work is part of the stated review framework. You should avoid generic statements like “we support equity” and instead explain actual actions:

• multilingual support,
• accessibility modes,
• diverse user testing,
• bias checks or prompt evaluation methods,
• policies for human review when AI outputs are uncertain.

4) Does the team have execution capacity for what you request?

A large award can be tempting, but judges evaluate feasibility. If you request significant funds without pilot capacity, data access, or clear milestones, you usually lose points. It is better to show a realistic path with clearly staged outputs than a grand blueprint.

Tracks and how they differ in practice

The official page lists three tracks:

• Augment Teaching: AI to support teachers, especially around student relationships and practical workflows.
• Augment Learning: AI to improve learner participation, including support for students with disabilities or learning differences.
• Augment Career Opportunities: AI for skill-building, mentorship, and pathways to meaningful work.

When choosing a track, don’t do it based on your favorite buzzword. Do it based on where your evidence will be strongest. If your prototype depends on teacher behavior change and professional development, pick Teaching even if there is a clear learner benefit. If your strongest outcome is confidence and participation among learners, pick Learning. If your product is strongest for portfolios, apprenticeships, project readiness, or mentorship loops, pick Career Opportunities.

What the program offers (beyond money)

The official page and associated Stanford posts describe several non-financial supports:

• Mentorship from Stanford community members across disciplines.
• Workshop-style interactions on learning science and teacher co-design.
• Visibility at the AI+Education Summit.
• Potential invitation to continue development in a summer cohort.

For practical applicants, this matters because these channels can open partnerships you may not be able to build independently, especially in schools, nonprofit networks, and research environments.

Eligibility: what was verified for this cycle

The official program page lists educators, researchers, technologists, nonprofits, and students. Stanford’s application announcement adds designers and entrepreneurs, says that both early-stage and existing projects could apply, and confirms the following cycle-specific details:

• At least one team member needed to be a current Stanford student, scholar, staff member, or alum.
• International applicants were eligible, subject to select country restrictions.
• Both early-stage ideas and existing projects ready to pilot, scale, or study were welcome.

The current official page does not state a general applicant age requirement. That claim has therefore been removed from this page. Because the form is now closed and no longer exposes its intake fields, treat the verified rules above as historical requirements for the 2026 call rather than live guidance for a future round.

If you use this framework for future rounds, confirm any requirements from the live page before spending application time.

Application process and current status

The 2026 process is no longer open, and the live form no longer exposes its questions. The verified process information is therefore limited to the official call description and Stanford’s application announcement. In that cycle, teams should have followed these steps:

1. Form a qualifying team. Include at least one current Stanford student, scholar, staff member, or alum. International applicants could participate subject to country restrictions.
2. Choose one track. Select Augment Teaching, Augment Learning, or Augment Career Opportunities based on the project’s main user and intended outcome.
3. Define the project. Explain the problem, the people affected, the role of AI, and how the design would augment human capability rather than remove important relationships or judgment.
4. Prepare for the scorecard. Address innovation, learning impact, fairness and inclusion, learning sciences and design, measurement, feasibility, and sustainability.
5. Submit through the official application form. The final deadline was January 12, 2026 at noon Pacific Time. The form is now closed, so no submission can be made for this round.

Stanford’s public archive does not currently provide a verified list of every form field or required attachment. This page therefore does not invent a document list, video requirement, budget template, or word limit. Anyone preparing for a future round should use only the new official form and instructions when they are published.

What a strong proposal would need to include

Even without current submission mechanics, reviewers still score in the same broad patterns: innovation, impact, fairness, learning science grounding, measurement, and feasibility.

To prepare in a way that transferably passes these criteria:

• Start with one explicit problem statement in one sentence.
• Explain who is affected and why now.
• State how your solution augments human work, not replaces it.
• Describe the intervention in practical terms: what happens in week one, month one, and month three.
• Define at least two outcomes and one method of measurement.
• Describe risks and mitigations (data handling, staff time, inequitable effects, deployment constraints).
• Show budget alignment with outcomes: every requested dollar should move one measurable outcome forward.

For teams, this is where people often fail: they spend too much energy on technology description and too little on context, workflow integration, and evidence collection.

Practical application preparation (historical guidance for future rounds)

Because the 2026 form is closed, what you can still do is prepare using a robust application template that mirrors likely expectations.

Step 1: Clarify your story

Draft a short narrative that covers three lines:

• Who is the user?
• What human burden exists today?
• How does AI change that burden?

Reviewers usually read fast, so this should be intelligible to someone outside your subfield.

Step 2: Make the outcome measurable

Create a tiny scorecard of outcomes and metrics before your technical plan:

• target group,
• target effect,
• baseline method,
• endpoint at 4–12 weeks,
• what success looks like.

Use conservative numbers and explain what would constitute failure. That builds credibility.

Step 3: Build an ethical and inclusion layer into the design

Include a dedicated section for fairness and access:

• who might be harmed by over-automation,
• what control options users have,
• how non-English and differently-abled users are supported,
• what your review and error-handling loops look like.

Step 4: Design an implementation plan that fits your resources

A review panel may accept that your idea is compelling but not likely if implementation is not realistic. Include staffing assumptions, pilot setting assumptions, and clear dependencies (data access, partner schools, legal approvals).

Step 5: Prepare concise supporting materials

The current archive does not verify which supporting formats the closed form requested. For a comparable future call, a small set of well-structured artifacts can still help reviewers assess execution: a concise problem brief, a measurement plan, a realistic implementation sequence, and a budget connected to outcomes. Treat those as preparation advice, not as a confirmed 2026 submission requirement.

Timeline model for deciding readiness

Even though this call is closed, this timeline is still useful for planning your next application.

• 6 to 8 weeks before a target deadline: align team and choose one track.
• 4 to 5 weeks before: finalize problem, user journey, and outcomes.
• 3 weeks before: produce draft deck/brief and run external review.
• 2 weeks before: film, record, and tighten visuals.
• 1 week before: verify all links, permissions, and team confirmation.
• 3 days before: final internal review and fallback edits.

In this challenge style, technical polish helps, but strategic clarity matters more than aesthetic perfection.

Common mistakes that weaken submissions

• Submitting a broad idea with no operational setting.
• Ignoring equity and inclusion details until the end.
• Confusing AI capability with educational impact.
• Designing around tool novelty rather than measurable outcomes.
• Underestimating teacher/workforce context and time constraints.
• Letting one technical founder dominate a team narrative without clear school or learner perspective.

Each of these is avoidable with structured planning. The reviewer can usually see through buzzwords quickly.

FAQ

Is this call still open?

No. The 2026 call ended on January 12, 2026 at noon Pacific Time. Stanford’s current official page lists awardees and does not announce a next cycle.

Can international teams apply?

Yes, subject to country restrictions. Stanford’s application announcement described the call as open to international applicants. Future rounds may use different rules.

Do non-Stanford teams have any pathway in?

For this 2026 round, at least one team member needed to be a current Stanford student, scholar, staff member, or alum. That was a cycle-specific requirement, not a promise about any future call.

What was the value beyond cash?

The official page lists $400,000 in total awards, mentorship and connections, learning-science and teacher co-design workshops, visibility at the AI+Education Summit on February 10 and 11, 2026, and a potential invitation for some projects to continue development in a summer 2026 cohort.

What should I do now if my idea fits?

Do not discard it. Turn your draft into a reusable applicant package:

• one-page problem statement,
• user outcome metrics,
• safety and inclusion notes,
• simple budget mapping.

Keep it ready for the next similar round or a sister program with similar criteria.

Official link and follow-up action

• Create+AI Challenge official page

Use the official Stanford page as the source of truth. Keep a one-page outcome brief, a practical implementation plan, and a budget tied to measurable milestones ready, but wait for a new official announcement before treating this program as open.