About the Fellowship
The Cambridge ERA:AI Fellowship is a 10-week research programme hosted in Cambridge, UK, with our Summer fellowship starting on 6th July, 2026.
What we provide
Full funding: Fellows receive a salary equivalent to £34,125 per year, which will be prorated to the duration of the Fellowship. On top of this, our fellows receive complimentary accommodation, meal provisions during working hours, visa support, and travel expense coverage.
Expert mentorship: Fellows will work closely with a mentor on their research agenda for the fellowship. See our Mentors page to learn about previous mentors (but note that ERA mentors differ for each cohort as we match fellows with mentors once they are accepted).
Research Support: Many of our alumni have gone on to publish their research in top journals and conferences such as NeurIPS. We provide dedicated research management support to help our fellows become strong researchers / policymakers in the field.
Community: Fellows are immersed in a living-learning environment. They will have a dedicated desk space at our office in Cambridge and are housed together. There are regular social events and community activities throughout the fellowship.
Networking and learning opportunities: We assist fellows in developing the necessary skills, expertise, and networks to thrive in an AI safety or policy career. We can facilitate introductions to many organisations in the field. In special cases, we also provide extra financial assistance to support impactful career transitions.
Application Process
Applications for our Summer 2026 Fellowship have now closed. If you'd like to hear about future cycles, you can express your interest and/or sign up to our mailing list below.
-
The first stage of the application process primarily consists of short essay questions (and should take you about 2 hours).
-
We will then invite the top ~20% of the candidates for stage 1 interview. These will happen mid-April.
-
The applicants that we are most excited about will then be invited to a shorter round 2 interview. When we invite you for this interview, we will also reach out to your references.
These will happen late-April. -
After the final acceptance offers have been made, we will work closely with the cohort of ERA ’26 fellows to develop your project ideas and match you with mentors.
Carson Ezell, Summer Research Fellow 2024
“This programme allowed me to openly and autonomously explore my ideas while providing significant support, minimising bureaucracy and limitations, and creating a conducive environment for research. I do not think an opportunity like this exists elsewhere.”
Who can apply?
Anyone! We are a talent-first programme and care about much more than just credentials. In fact, we are excited to support fellows from a wide range of subject areas who are committed to our mission. There are no formal eligibility restrictions beyond being 18 or older.
The Cambridge ERA:AI Fellowship is especially (though not exclusively!) valuable for:
Researchers at any career stage looking to apply their expertise to high-priority questions in AI safety or governance, explore a new research direction, or work across disciplinary boundaries;
Professionals from technical, policy, security, legal, economic or other relevant fields who want to bring their existing expertise to frontier AI safety and governance;
People seeking to deepen or redirect an existing research agenda towards problems arising from increasingly capable AI systems; and
People with distinctive domain expertise or perspectives that could improve how we understand and address risks from advanced AI.
Our Research
During the Fellowship, ERA fellows develop and complete a research project on technical and/or governance measures to mitigate the risks posed by frontier AI systems. They have the support of ERA’s research managers, mentors, and wider institutional network. For examples of previous projects, see our Previous Research Projects page.
At ERA, we support research across the full stack of frontier AI governance & technical AI safety, mentored by researchers from a wide range of organisations across government, academia & industry, including University of Cambridge, UK AI Security Institute, Centre for the Governance of AI, RAND Corporation, Google DeepMind, OpenAI, Anthropic & more.
See below and here for some further detail on the three streams at ERA.
01
Technical
As AI systems become more capable and more autonomous, understanding how they behave, identifying dangerous capabilities, and maintaining meaningful human oversight all become increasingly important.
At ERA, we support research across evaluations, interpretability, robustness, alignment and control. Our fellows work on questions ranging from measuring and eliciting dangerous capabilities, to understanding model internals, designing robust safeguards, monitoring model behaviour, and developing techniques for controlling systems whose objectives may not be fully aligned with those of their operators. In doing so, they contribute directly to some of the central research and engineering challenges in frontier AI safety.
Fellows in this stream pursue research to ensure that that increasingly capable AI systems can be controlled via human oversight, and also that advanced AI systems are both built with appropriate safeguards to avoid harmful or unpredictable behaviour. We are also interested in emerging challenges such as detecting deceptive or strategically evasive behaviour, controlling systems capable of automating parts of AI R&D, accelerating alignment research across the full stack of LLM training, and developing safeguards which remain effective under increasingly capable models.
02
Governance
As frontier AI systems become more capable, decisions about how they are developed, deployed and governed become increasingly consequential. Effective governance requires institutions that can understand rapidly changing risks, establish credible rules and safeguards, and coordinate internationally where risks and incentives cross national borders.
At ERA, we support research which advances international cooperation, strengthens regulatory tools in key jurisdictions, and improves governance and accountability within frontier AI companies. Our fellows work on questions spanning regulatory design and implementation, frontier safety frameworks, corporate governance and international coordination, helping build the institutional capacity needed to govern increasingly powerful AI systems.
Fellows in this stream will pursue research to ensure that the institutions governing advanced AI can keep pace with the systems themselves: adapting as capabilities and risks change, creating meaningful accountability for frontier developers, and enabling effective coordination between governments and companies. We are also interested in emerging governance challenges such as for increasingly autonomous AI agents and policy options for building the security & verification backbone for frontier AI.
03
Technical AI Governance
As frontier AI systems become more capable, effective governance increasingly depends on technical facts and infrastructure: what systems can do, how much compute they use, where and how they are deployed, and whether claims about their development can be independently verified. Laws are difficult to implement if governments and other actors cannot reliably measure capabilities, monitor deployed systems, or determine whether requirements are actually being followed. Technical AI governance asks how the technical substrate of AI can make governance more informed, precise and enforceable.
At ERA, we support research across compute governance, auditing, monitoring, hardware-enabled verification, and technical standards. Our fellows work on questions such as how regulators can identify systems that cross risk-relevant thresholds, how hardware and compute providers can enable privacy-preserving verification, and how increasingly autonomous AI systems can be monitored and audited. Their work helps translate governance objectives into technical mechanisms that can function in real systems and inform policy design by working closely with government organisations such as AISIs, frontier AI labs and policymakers.
Our goal for this stream is to ensure that the governance of advanced AI does not rely solely on trust or self-reporting, but can be grounded in credible mechanisms for measurement, verification and enforcement. We are particularly interested in technical tools that give governments and other relevant actors better visibility into frontier AI development, allow important claims or commitments to be independently verified, and enable meaningful constraints to be implemented while preserving privacy and security.