AI Safety Policy Entrepreneurship Fellowship 2026–2027: $5,000 Stipend, Policy Training and Frontier AI Governance Experience

The AI Safety Policy Entrepreneurship Fellowship (PEF) 2026–2027 offers an opportunity for early- to mid-career professionals who want to turn ideas about frontier artificial intelligence safety, security and governance into practical policy solutions.

Running from September 30, 2026, through February 28, 2027, the fellowship is designed around policy entrepreneurship rather than academic research alone. Fellows will work part-time as affiliates of the Federation of American Scientists (FAS), developing a concrete policy idea, engaging decision-makers, building coalitions and identifying realistic pathways for their proposal to influence policy.

The fellowship provides a $5,000 stipend, with an opportunity for an additional $1,000 merit award for the Fellow who develops the most specific and actionable policy memo and advocacy plan. Selected travel expenses for required in-person events are also supported through an additional travel stipend.

Applications must be submitted by 11:59 PM ET on September 7, 2026.

What Is the AI Safety Policy Entrepreneurship Fellowship?

The Policy Entrepreneurship Fellowship is an initiative designed to help participants move promising policy ideas beyond the conceptual stage and toward real-world implementation.

Rather than concentrating exclusively on producing research papers, the fellowship emphasizes the practical process of making policy happen.

Participants will be encouraged to:

• Identify an important policy problem.
• Develop a practical and implementable solution.
• Identify institutions capable of implementing the solution.
• Understand relevant decision-makers.
• Engage policymakers and other stakeholders.
• Build coalitions around their proposal.
• Identify opportunities or “policy windows” for action.
• Develop an advocacy strategy.
• Produce a policy memo.
• Move their idea toward practical policy adoption.

For the 2026–2027 cohort, the fellowship will specifically focus on frontier AI safety issues.

This makes the program particularly relevant to professionals interested in the intersection of artificial intelligence, public policy, national security, cybersecurity, governance, technology regulation and catastrophic risk management.

Fellowship Focus: Frontier AI Safety

The fellowship will explore policy challenges arising from increasingly capable and advanced AI systems.

Fellows are encouraged to examine different dimensions of frontier AI safety, including technical, regulatory, institutional and governance challenges.

Technical AI Safety

One area of focus is ensuring that frontier AI systems are:

• Reliable.
• Predictable.
• Secure.
• Aligned with human values.
• Designed and deployed in ways that reduce serious risks.

Applicants with technical AI expertise may therefore be able to translate their technical knowledge into policy proposals addressing the governance of advanced AI systems.

AI Governance and Policy

Another major area involves developing frameworks for governing the development and use of frontier AI.

Potential questions may include:

• What policies should govern advanced AI development?
• Which institutions should oversee high-risk AI systems?
• How should governments establish appropriate safeguards?
• What regulatory mechanisms could reduce catastrophic AI risks?
• How can governments build the capacity required to govern rapidly developing technologies?

Risk Assessment and Mitigation

Fellows may also explore ways to identify, evaluate and reduce risks associated with increasingly capable AI systems.

Potential policy approaches could involve:

• Risk assessment frameworks.
• Safety requirements.
• Risk classification systems.
• Governance mechanisms.
• Monitoring and oversight.
• Mitigation standards.
• Institutional preparedness.

Transparency and Accountability

The fellowship also welcomes ideas addressing transparency and accountability in AI development and deployment.

This could include examining how governments, companies and other institutions can improve understanding of:

• How advanced AI systems are developed.
• What capabilities they possess.
• What risks they create.
• How they are deployed.
• Who is responsible for managing those risks.
• How oversight can be strengthened.

Priority Policy Themes

The fellowship is particularly interested in policy proposals addressing several emerging frontier AI safety and governance issues.

These include:

Cybersecurity

Applicants could examine cybersecurity risks associated with advanced AI systems and develop policy mechanisms designed to strengthen security.

Capability Thresholds

Another potential area is determining whether specific AI capabilities should trigger additional regulatory, safety or oversight requirements.

Risk Taxonomies

Fellows may explore systems for categorizing different types and levels of AI risk so that policymakers can develop proportionate responses.

Critical Infrastructure

Advanced AI could have implications for critical infrastructure and essential systems. Policy proposals could examine how these risks should be assessed and mitigated.

Compute Governance

Computational resources play an important role in the development of frontier AI. Fellows may explore governance mechanisms relating to access to, monitoring of or oversight of advanced computing resources.

Model Proliferation

The fellowship is interested in questions surrounding the proliferation of AI models, including the policy implications of open versus closed systems.

Policy Design and Government Capacity

Applicants may also propose broader approaches to policy design, institutional arrangements and government capacity for managing catastrophic risks associated with advanced AI.

What Is Expected From Fellows?

The fellowship is designed to give participants flexibility to pursue a policy project that they believe can create meaningful impact.

However, Fellows are expected to develop a concrete policy proposal rather than simply explore a broad research topic.

The central objective is to identify a specific frontier AI safety problem and develop a feasible pathway toward policy adoption.

Possible mechanisms for implementation could include:

• Legislation.
• Regulation.
• Government standards.
• Procurement requirements.
• Institutional design.
• Government policies.
• Governance frameworks.
• Other practical policy mechanisms.

Key Fellowship Deliverable: Policy Memo

One of the central responsibilities will be to develop and publish a policy memo addressing a specific frontier AI safety challenge.

The memo should go beyond describing the problem.

It should include:

1. A clearly defined policy challenge.
2. Evidence and analysis demonstrating why the issue matters.
3. A practical policy solution.
4. The institutions and decision-makers relevant to implementation.
5. Key stakeholders who need to be engaged.
6. A strategy for moving the proposal toward adoption.
7. Consideration of practical barriers to implementation.

The goal is to produce a policy document that can contribute to an actual policy conversation.

Policy Entrepreneurship Training

Fellows will participate in training sessions designed to strengthen their ability to turn ideas into policy impact.

Training will address areas such as:

• Frontier AI safety.
• Policymaking processes.
• Policy entrepreneurship.
• Stakeholder engagement.
• Translating technical expertise into policy.
• Identifying policy opportunities.
• Building coalitions.
• Engaging decision-makers.
• Moving policy proposals toward implementation.

Participants will also take part in reading groups intended to deepen their understanding of frontier AI safety and practical policymaking.

In-Person California Retreat

The fellowship includes an in-person California retreat tentatively scheduled for:

November 4–7, 2026

The retreat will focus on:

• Mentorship.
• Community building.
• Developing fellowship projects.
• Advancing policy proposals.
• Connecting Fellows with relevant expertise and perspectives.

An additional travel stipend will be available to cover selected expenses associated with the California retreat.

Washington, DC Capstone Event

Fellows are also expected to participate in an in-person capstone event in Washington, DC, tentatively scheduled for the week of:

February 22, 2027

The capstone event is intended to provide an opportunity for Fellows to:

• Launch their policy memos.
• Disseminate their policy recommendations.
• Engage policymakers.
• Connect with experts.
• Engage relevant stakeholders.
• Present the outcomes of their fellowship projects.

This component gives Fellows an opportunity to move their work beyond the fellowship itself and into conversations with people involved in policymaking and governance.

Time Commitment

The fellowship is structured as a part-time opportunity.

Fellows are expected to dedicate an average of:

5 hours per week, or approximately 20 hours per month

to the fellowship during the period from September 30, 2026, through February 28, 2027.

This time commitment includes mandatory fellowship programming.

Participants will also take part in twice-monthly group check-ins.

FAS will make reasonable efforts to schedule these sessions at times that accommodate participants.

Select absences may be accommodated when participants provide appropriate notice, including absences related to:

• Illness.
• Work conflicts.
• Vacation.

Exceptional circumstances may also be considered when determining whether a participant can be exempted from certain mandatory sessions.

Individual Mentorship

Each Fellow will receive individual support from FAS.

The fellowship includes:

• Monthly individual check-ins with a dedicated FAS staff mentor.
• Additional meetings when needed.
• Additional communication and support as projects develop.
• Assistance connecting Fellows with relevant experts and policy actors.

This mentorship component is particularly valuable for applicants who have strong technical or subject-matter expertise but limited experience navigating policy processes.

Fellowship Benefits

The AI Safety Policy Entrepreneurship Fellowship provides several professional and financial benefits.

$5,000 Fellowship Stipend

Each Fellow will receive a:

$5,000 stipend

This provides financial support for participants during the fellowship period.

Additional $1,000 Merit Award

An additional:

$1,000 merit award

may be provided to the Fellow whose policy memo and advocacy plan are judged to be the most specific and actionable.

This means that participants have an opportunity to receive up to $6,000 in total financial awards, excluding eligible travel support.

Travel Support

An additional travel stipend will help cover selected expenses associated with:

• The California retreat.
• The Washington, DC capstone event.

Professional Editing

Fellows will receive professional editing support for written products developed during the fellowship.

This can help improve the clarity, structure and professional quality of policy-related writing.

Policy Engagement Training

Participants will receive training from FAS staff focused on practical policy entrepreneurship.

This can help Fellows understand how to move from research and ideas toward policy influence.

Access to FAS Programming

Fellows will receive invitations to relevant FAS programming.

They will also have access to curated updates concerning the rapidly evolving federal science and technology policy landscape.

Deep-Dive Sessions

Participants will take part in interactive sessions examining practical aspects of policy entrepreneurship.

These sessions can help Fellows better understand how policy processes work in practice.

Expert Connections

FAS will support Fellows in connecting with relevant experts and policy actors.

This can help participants expand their professional networks and identify individuals and institutions that could contribute to their policy projects.

Who Is the Fellowship Designed For?

The fellowship is particularly well suited to early- to mid-career professionals who have strong subject-matter expertise but limited direct experience working in public policy.

The program is intended for people who want to learn how to translate their knowledge into practical policy impact.

Potential applicant backgrounds include:

• Technical AI research.
• Artificial intelligence.
• Academia.
• Think tanks.
• Civil society.
• Technology industry.
• Law.
• Cybersecurity.
• National security.
• Public policy.
• Science and technology policy.
• Related technical or policy disciplines.

The fellowship is not limited to people who have previously worked as professional policymakers.

In fact, candidates with substantial expertise in another field who want to learn how to translate that expertise into policy action may be particularly well suited to the program.

Selection Criteria

Applications will be assessed holistically.

The selection process will consider the overall strength of the applicant and their proposed policy project rather than relying on a single qualification.

1. Understanding of a Significant AI Safety Challenge

Applicants should demonstrate a clear understanding of an important frontier AI safety, security or governance problem.

The problem should be meaningful and sufficiently specific to support a practical policy intervention.

2. Concrete Policy Idea

Applicants should propose a concrete policy idea rather than simply identifying a broad concern about AI.

The proposal should be:

• Specific.
• Practical.
• Implementation-oriented.
• Connected to a real policy challenge.
• Supported by a credible pathway toward adoption.

3. Understanding of Key Stakeholders

Applicants should consider which institutions, decision-makers and stakeholders would be necessary to implement their proposal.

This demonstrates that the applicant understands policy as a process involving multiple actors rather than simply producing a recommendation.

4. Understanding of Practical Policy Challenges

Strong candidates should recognize that good policy ideas can face significant implementation barriers.

Applicants should demonstrate willingness to think constructively about:

• Political constraints.
• Institutional limitations.
• Stakeholder interests.
• Implementation challenges.
• Government capacity.
• Competing priorities.
• Practical pathways toward adoption.

5. Curiosity and Initiative

The fellowship seeks people who demonstrate curiosity, initiative and genuine interest in translating ideas into practical impact.

Applicants should demonstrate that they are prepared to take initiative rather than simply conduct research.

What Makes a Strong Application?

A competitive application should clearly connect problem, solution and implementation.

A strong proposal should answer questions such as:

1. What specific frontier AI safety problem are you addressing?
2. Why is this problem important?
3. Who is affected by the problem?
4. What policy intervention are you proposing?
5. Which institution could implement it?
6. Who are the relevant decision-makers?
7. Which stakeholders would support or oppose it?
8. What barriers could prevent implementation?
9. How could those barriers be addressed?
10. What practical opportunity exists for advancing the proposal?

The strongest applications are therefore likely to demonstrate not only knowledge of AI safety but also an understanding of how policy change actually occurs.

Why the Fellowship Is Valuable for AI Policy Careers

The rapid development of frontier AI is creating increasing demand for professionals who understand both technology and public policy.

However, technical expertise alone does not necessarily translate into policy influence.

This fellowship addresses that gap by helping participants develop the skills required to move from:

Technical or subject-matter expertise → Policy idea → Stakeholder engagement → Advocacy → Policy adoption

Participants can therefore develop capabilities in:

• Policy analysis.
• Policy writing.
• Stakeholder mapping.
• Coalition building.
• Decision-maker engagement.
• Advocacy strategy.
• Policy communication.
• Governance.
• AI safety.
• Science and technology policy.

For professionals considering long-term careers in AI governance, technology policy, national security, cybersecurity or public-interest technology, these capabilities can be particularly valuable.

Click HERE to read more.

Fellowship Timeline

The fellowship runs for approximately five months.

Fellowship begins: September 30, 2026

California retreat: November 4–7, 2026, tentatively

Washington, DC capstone: Week of February 22, 2027, tentatively

Fellowship concludes: February 28, 2027

Application deadline: September 7, 2026, at 11:59 PM ET

Application Deadline and Submission Requirements

Applications must be submitted by:

11:59 PM ET on September 7, 2026

Applicants should be aware that applications are accepted only through the designated application platform.

When applicants select the application link, they will be redirected to the relevant platform and will need to create an account in order to submit their application.

Candidates should therefore avoid waiting until the final hours to begin the application process.

APPLY NOW!

Important Application Considerations

Before applying, candidates should make sure their proposed project is sufficiently focused.

A broad proposal such as “AI safety should be improved” is unlikely to demonstrate the level of specificity expected from a policy entrepreneurship fellowship.

Instead, applicants should identify a concrete problem and explain how a particular policy mechanism could address it.

For example, a proposal might focus on:

• A specific cybersecurity risk associated with frontier AI.
• A governance mechanism for AI capability thresholds.
• A framework for categorizing advanced AI risks.
• AI risks to critical infrastructure.
• Governance of computing resources.
• Oversight of AI model proliferation.
• Standards for high-risk AI systems.
• Institutional capacity for managing catastrophic AI risks.

The proposal should then connect the problem to a realistic implementation pathway.

Conclusion

The AI Safety Policy Entrepreneurship Fellowship 2026–2027 offers an opportunity for professionals to develop practical solutions to some of the emerging governance challenges associated with frontier artificial intelligence.

The fellowship is particularly distinctive because it goes beyond traditional research. Participants are expected to identify a concrete policy problem, develop an actionable solution, understand the institutions and stakeholders involved, build relationships with decision-makers and pursue a realistic pathway toward implementation.

With a $5,000 stipend, potential $1,000 merit award, travel support, professional editing, policy training, mentorship and access to experts, the program provides both financial and professional support for participants seeking to develop careers at the intersection of AI safety and public policy.

Professionals from technical AI research, academia, cybersecurity, law, national security, civil society, industry, think tanks and related fields are encouraged to consider applying, particularly if they have a strong policy idea but limited direct experience navigating policymaking.

The application deadline is September 7, 2026, at 11:59 PM ET. Candidates should focus on presenting a specific frontier AI safety challenge, a credible policy solution and a practical strategy for moving that solution toward real-world adoption.

Click HERE for more fellowship opportunities.