Mila AI Policy Fellowship

This thematic area explores how to develop, deploy, and regulate AI systems in ways that are safe, rights-respecting, and socially sustainable. As AI becomes more powerful, general-purpose, and autonomous—particularly through foundation models, multi-agent systems, and open-source deployments—safety concerns expand beyond technical robustness to include long-term risk management and global coordination. This area considers both technical and institutional safeguards needed to mitigate misuse, prevent harm, and preserve democratic accountability as AI scales in complexity and impact. Fellows will examine how safety, privacy, transparency, and human rights can be built into AI systems and governance frameworks across the entire AI lifecycle.

Example topics include:

• Improved evaluation of foundation models and other advanced systems to understand their limitations, risks, and potential harms across high-impact domains such as healthcare, law enforcement, or education
• Effective monitoring, auditing, and remedy mechanisms to respond to harms caused by AI systems, including those deployed in high-stakes or public-facing environments
• Appropriate policy responses to emerging technical challenges, such as adversarial robustness, privacy-preserving techniques, and machine unlearning
• National and global governance strategies to address safety risks associated with open-source AI models and multi-agent systems
• institutional capacity-building by identifying tools, safeguards, and coordination mechanisms to anticipate, mitigate, and respond to systemic AI risks

This thematic area explores how to ensure inclusive, equitable, and effective adoption of AI technologies across sectors to boost productivity, foster innovation, and strengthen competitiveness. While new AI capabilities—such as enterprise-level agents, automated planning systems, and multimodal tools—are transforming how work is organized and how value is created, adoption remains uneven. Structural barriers such as high costs, skills shortages, limited infrastructure, and regulatory uncertainty continue to hinder uptake, especially among small and medium-sized enterprises (SMEs) and under-resourced public institutions.

Example topics include:

• Sector-specific policies to lower barriers to AI access and adoption, particularly for SMEs and public sector actors.
• Strategies for workforce development and reskilling to address gaps in AI-related technical and applied capabilities.
• Governance frameworks to ensure trustworthy and responsible AI adoption in areas such as education, healthcare, and local governance.
• Incentives, funding mechanisms, and public-private partnerships to accelerate inclusive and sustainable adoption of productivity-enhancing AI tools

The rise of generative AI, personalized recommendation systems, and multimodal models is transforming the media landscape and reshaping how people access, engage with, and trust information. This thematic area explores the governance of AI-enabled media technologies and their far-reaching implications for public trust, electoral integrity, and the resilience of democratic systems. It also considers how these tools are affecting journalistic practices, content moderation, and the growing risks of ideological segregation.

Example topics include:

• Disinformation and the integrity and sovereignty of digital information ecosystems
• Evaluating the policy implications of real-time deepfake generation and detection technologies
• Governance strategies for large language models used in automated news production
• Transparency requirements for AI-curated content in political campaigns or civic platforms

Growing inequalities and a lack of diversity in AI risks overlooking the specific needs, local expertise and cultural contexts of historically marginalised and excluded groups. This thematic area explores policy implications and solutions to mitigate these disparities, to prevent harm and to ensure an inclusive, beneficial, and rights-based development of AI systems and processes that supports economic and social development. Substantive equality in AI builds on GPAI recommendations for Substantive Equality in AI, the OECD AI principles, the UNESCO Recommendation on the Ethics on AI, the SDGs, the Global Digital Compact and other work advancing responsible AI and focuses on removing constraints and enhancing capabilities to ensure the enjoyment of human rights within and throughout AI ecosystems and related policy-making.

Example topics include:

• Privacy and anonymity considerations in body scanning technologies, gender feature extraction in biometrics tasks, and other AI systems and processes
• Equity metrics and measurements and effective model evaluations and auditing procedures
• Contextual liability for non-discrimination in AI systems in proportion to other accountability measures such as level of transparency, interpretability, and explainability
• Effective transparency and accountability measures for harm prevention and effective access to justice, for example in relation to AI-related technology-facilitated gender-based violence (TFGBV)
• International AI benefitsharing and Global Majority participation in international AI governance processes

This thematic area explores how to align AI development with global sustainability priorities, using AI applications to be a driver of climate mitigation and adaptation. At the same time as the widespread adoption of AI accelerates, so too does its environmental impact—driven by the growing energy and water demands of data centers and the resource intensity of training and deploying large models. This area considers both the environmental footprint of AI systems and their potential to support climate action, improve resource management, and enable sustainable innovation across sectors.

Example topics include:

• Applications of AI and climate, including climate modelling and risk forecasting , materials discovery for energy systems, grid optimization and energy use, biodiversity preservation and forest management.
• Improving assessments of AI’s environmental footprint, including direct energy and water usage and broader systemic effects across sectors.
• Policies to enhance the adoption of AI applications for climate, address compute-related regional disparities and promote energy-efficient AI infrastructure, such as advances in hardware, cooling technologies, and Small Language Models.

This thematic area explores how AI systems, deployment and governance can respect and reinforce Indigenous rights, data sovereignty, and self-determination. As AI increasingly influences governance, economic development, and knowledge systems, there is an urgent need to ensure that Indigenous Peoples—particularly in the Canadian context—are not only protected from harm but are also positioned as leaders and decision-makers in shaping AI policy, innovation, and infrastructure. This area considers strategies to support Indigenous authority over data, language, and digital tools while advancing inclusive participation and community capacity building in the AI-driven economy.

Example topics include:

• Indigenous language revitalization in and through AI, ensuring culturally grounded and ethically sound approaches aligned with community priorities and international frameworks such as the UN Declaration on the Rights of Indigenous Peoples
• Embedding Indigenous data sovereignty into AI governance frameworks at federal, provincial, and institutional levels, including through legal, policy, and technical mechanisms
• Strengthening Indigenous control over data governance in AI, ensuring Indigenous nations have authority over how their data is collected, stored, and used.
• Addressing collective data privacy, consent, and control in machine learning systems affecting Indigenous communities
• Exploring ethical approaches to AI adoption and impact mitigation in Indigenous communities’ contexts
• Strengthening Indigenous inclusion in the digital and AI economy through talent development, entrepreneurship pathways, and investments in digital infrastructure

The adoption of AI in the public sector holds significant potential to improve governance, enhance service delivery, and optimize decision-making. However, governments face unique challenges in implementing AI responsibly, including concerns over ethics, transparency, workforce readiness, and public trust. While AI can streamline administrative processes, detect fraud, and support policy analysis, its misuse or lack of oversight could exacerbate biases, reduce accountability, and undermine citizen rights. This theme addresses how public-sector AI—including automated decision systems and recommendation tools—can be deployed and governed transparently and equitably.

Example topics include:

• Auditing the use of AI in eligibility systems for services like immigration, housing, or social benefits
• Identifying safeguards to protect privacy, prevent discrimination, and maintain meaningful human oversight in automated public decision-making
• Designing accountability frameworks for public-facing AI agents in local governance and service provision
• trustworthy, interoperable, and vendor-neutral AI systems
• Strategies for building public sector AI capacity, including civil servant training, in-house technical expertise, and reduced reliance on external vendors

Junior to senior researchers and professionals with interdisciplinary backgrounds who are already active in academia, civil society, public service, or the private sector, and who seek to deepen their engagement with AI policy.

To be eligible, applicants must:

• Hold a graduate degree (M.Sc., M.A., or equivalent) in a relevant field such as public policy, law, ethics, AI, sociology, economics, related disciplines or list other equivalent experience.
• Have a minimum of three years of relevant academic or professional experience.
• Demonstrate experience with interdisciplinary or multi-stakeholder initiatives (an asset).
• Submit a complete and original project proposal feasible within a six-month timeline.

We welcome policy-relevant and interdisciplinary projects aligned with one or more of the following thematic areas: AI Safety and Governance, AI Adoption and Productivity, AI, Media, and Democracy, Substantive Equality in AI, AI and Climate, Indigenous AI and AI in the Public Sector.

Your project proposal should demonstrate a clear, well-developed, and feasible plan that aligns with the objectives of the fellowship. It must include the following components:

• Project Title (Max. 15 words): A clear, descriptive title reflecting the main focus of your project.
• Project Summary (Max. 100 words): A brief overview of your project, including the central topic, key research or policy question(s), and its relevance to AI policy.
• Project Rationale and Objectives (Max. 350 words): Describe the motivation behind your project, including main objectives, target audience, and intended policy impact. Identify specific policy issue(s) or gap(s) it will address and the guiding questions. Indicate which fellowship theme(s) your project aligns with, or explain the relevance if your topic falls outside the predefined areas. If applicable, mention: Any Mila researcher you intend to collaborate with and why (no prior connection is required) and any additional collaborators or sponsorships.
• Methodology and Approach (Max. 250 words): Describe your interdisciplinary approach and the key methods or frameworks. Indicate whether your project builds on existing research or proposes a new direction.
• Feasibility and Timeline (Max. 100 words): Outline how you will complete the project over the six-month part-time fellowship. Identify major phases, milestones, or planned activities.
• Suitability for the Fellowship (Max. 150 words): Explain why you are well-positioned to carry out this project. Highlight relevant academic, professional, or lived experience and your motivation. Describe your experience with and commitment to interdisciplinary exchange and collaboration, including examples from past work across disciplines, sectors, or communities.

Yes. Applicants may submit proposals outside the predefined areas, but should clearly explain the relevance to AI policy and why the topic is important to address through the Fellowship. Please indicate potential Mila or external collaborators.

Yes. The Fellowship will pair you with a Mila advisor. You may propose a collaboration with a Mila researcher in your application (though this is not required at the application stage). You can also include other collaborators, including funders or institutional partners.

Your application must include:

• A completed application form
• A project proposal (max. 1,000 words / 2 pages)
• A published writing sample
• Two references (please ensure you have contacted them prior to submitting)
• Your CV

Fellows are expected to complete:

• A 6–8 page policy brief
• A policy roundtable or stakeholder event
• A Mila community event
• A dissemination strategy
• A 10-page internal research summary report

Applications are reviewed by a selection committee composed of Mila staff and advisors. Selection is based on the quality and feasibility of the proposal, its relevance to AI policy, interdisciplinary orientation, match with a Mila advisor, and the applicant’s experience and motivation. Shortlisted candidates may be invited for a brief interview before final decisions are made.

The working languages of the fellowship are English and French.

No. The Fellowship is part-time, with a workload of 10–15 hours per week between September and February. Fellows are expected to remain anchored in their home institutions or professional roles.

Fellows will receive CAD 40 per hour, for up to 15 hours per week over the six-month program.

No. The program is open to international applicants. The Fellowship is delivered in a hybrid format, with in-person activities scheduled in Montreal, Quebec. Visa support can be provided for selected fellows who need it.

Yes. If you are selected and require a visa to attend the in-person component, Mila can support your application process.

No. There are no application fees. Participation is fully funded, including travel and event costs for the in-person component (up to 14 days). If you wish to extend your in-person stay during the fellowship period, you are welcome to do so at your own expense.

Yes. All travel and event-related costs associated with the required in-person components of the fellowship are covered by the program.

While we encourage live participation in key sessions, we aim to accommodate fellows in different time zones. Core sessions will be scheduled with input from the cohort to maximize inclusivity.