ERA Fellowship Winter 2027 in Cambridge, UK: Fully Funded 10-Week AI Safety and Governance Research Programme

Applications are now open for the ERA Fellowship Winter 2027, an intensive 10-week research programme in Cambridge, United Kingdom, designed for talented individuals interested in advancing research on technical AI safety, AI governance and technical AI governance.

The Winter 2027 ERA Fellowship will take place from 18 January to 26 March 2027 in Cambridge. Fellows will have the opportunity to work full-time on substantive research projects addressing some of the most important questions surrounding increasingly capable artificial intelligence systems.

ERA seeks researchers, professionals, PhD students and other talented individuals from around the world who want to contribute to understanding and mitigating potentially catastrophic risks associated with frontier AI. Importantly, applicants do not necessarily need to come from a traditional artificial intelligence research background. ERA specifically encourages applications from early-career researchers, PhD students and working professionals, including people from fields such as computer science, engineering, law and related disciplines.

Successful fellows will receive a competitive stipend, while ERA will also provide or cover key costs associated with participating in the programme, including lodging, transportation, visa support and meals during working hours.

Applications for the Winter 2027 programme must be submitted by 13 September 2026 at 11:59 PM Anywhere on Earth (AoE).

About ERA and Its Mission

ERA works to accelerate talent, advance research and strengthen organisations working to understand and reduce risks arising from increasingly capable frontier technologies, particularly advanced artificial intelligence.

Its broader mission focuses on advancing scientific and policy breakthroughs that can help society mitigate risks from frontier technologies.

In the context of artificial intelligence, ERA is concerned with complex challenges that could emerge as AI systems become increasingly capable and autonomous. These include questions surrounding the safe development and deployment of frontier AI, the preservation of meaningful human agency, governance mechanisms for powerful AI systems, and ways of preventing catastrophic outcomes associated with advanced artificial intelligence.

ERA’s work therefore brings together researchers working across both technical and governance disciplines.

Rather than treating AI safety solely as a technical engineering problem or solely as a policy challenge, the programme supports work across the broader landscape of frontier AI safety and governance.

What Is the ERA Fellowship Winter 2027?

The ERA Fellowship Winter 2027 is a 10-week research and entrepreneurship programme based in Cambridge, UK.

The programme will run from:

Start date: 18 January 2027
End date: 26 March 2027
Duration: 10 weeks
Location: Cambridge, United Kingdom
Application deadline: 13 September 2026 at 11:59 PM Anywhere on Earth

The fellowship is intended for researchers and entrepreneurs interested in understanding and mitigating catastrophic risks associated with frontier AI systems.

During the programme, fellows work full-time on research questions they consider important while receiving mentorship, research management support and access to a broader community working on AI safety and governance.

The fellowship also provides opportunities for projects that fall between established technical and policy research areas, making it particularly relevant for applicants interested in interdisciplinary approaches to advanced AI risks.

ERA Fellowship Winter 2027 Research Tracks

For the Winter 2027 cohort, ERA is accepting fellows across three main research tracks:

1. AI Governance
2. Technical AI Governance
3. Technical AI Safety

Applicants should carefully consider which track most closely matches their interests, experience and proposed research direction.

Some questions on the application form only become available after an applicant selects a research track.

1. Technical AI Safety

As artificial intelligence systems become increasingly capable and autonomous, understanding their behaviour and ensuring that humans can maintain meaningful oversight becomes more important.

The Technical AI Safety track supports research addressing some of the central scientific and engineering problems associated with frontier AI safety.

Research areas supported under this track can include:

• AI evaluations
• Interpretability
• Robustness
• AI alignment
• AI control
• Monitoring model behaviour
• Measuring potentially dangerous capabilities
• Eliciting capabilities from advanced AI systems
• Understanding the internal mechanisms of AI models
• Developing robust safeguards
• Controlling systems whose objectives may not be completely aligned with those of their operators

These areas address difficult questions about how advanced AI behaves, how dangerous capabilities can be identified and measured, and how powerful systems can remain subject to meaningful human oversight.

Fellows interested in machine learning, computer science, AI engineering, interpretability, alignment or related technical disciplines may find this track particularly relevant.

2. AI Governance

The AI Governance track focuses on the institutions, policies, regulatory mechanisms and international arrangements that may be necessary as frontier artificial intelligence becomes increasingly powerful.

As AI capabilities grow, decisions surrounding how advanced AI systems are developed, deployed and governed may have significant consequences.

Effective AI governance therefore requires institutions capable of understanding rapidly evolving risks, developing credible rules and safeguards, and coordinating across national borders where risks and incentives affect multiple jurisdictions.

ERA supports governance research in areas including:

• International cooperation on advanced AI
• AI regulatory tools
• Regulatory design and implementation
• Frontier AI safety frameworks
• Corporate governance
• Governance and accountability within frontier AI companies
• International coordination
• Institutional capacity for governing advanced AI systems

This track may be especially relevant to researchers and professionals with backgrounds in areas such as law, public policy, international relations, economics, political science, governance and related fields.

However, ERA does not restrict applications to these backgrounds.

3. Technical AI Governance

Technical AI Governance, sometimes described as TechGov, operates at the intersection of technology and public governance.

As frontier AI systems become increasingly capable, effective regulation may depend heavily on governments and institutions having access to reliable technical information and infrastructure.

For example, regulators may need mechanisms for determining whether an AI system has crossed a particular capability or risk threshold. Policymakers may also need reliable methods for monitoring deployed AI systems or verifying whether organisations are complying with particular requirements.

Technical AI governance explores how the technical foundations of artificial intelligence can make governance more informed, precise and enforceable.

Research areas can include:

• Compute governance
• AI auditing
• Monitoring
• Hardware-enabled verification
• Technical standards
• Risk-relevant AI capability thresholds
• Verification mechanisms
• Privacy-preserving verification
• Technical infrastructure supporting AI regulation

Projects may investigate questions such as how regulators could identify systems that exceed risk-relevant thresholds or how hardware and computing providers could facilitate effective verification while preserving privacy.

This makes the track particularly relevant for people interested in combining technical knowledge with governance, regulation or policy.

What Will ERA Fellows Do?

One of the central components of the fellowship is the development of a substantive research project.

During the 10-week programme, fellows work on their research with support from experienced mentors.

ERA’s mentor and institutional network extends across government, academia and industry and includes researchers connected with organisations and institutions such as the University of Cambridge, UK AI Security Institute, Centre for the Governance of AI, RAND Corporation, Google DeepMind, OpenAI and Anthropic, among others.

The fellowship is designed to give participants significant scope to develop research addressing important questions within their chosen area.

In addition to mentorship, fellows receive dedicated research management support and opportunities to discuss their research with other fellows and members of the Cambridge research community.

Build a Substantive AI Research Portfolio

A major objective of the ERA Fellowship is helping participants develop meaningful research experience and produce substantive work.

Rather than simply attending lectures or seminars, fellows spend the programme actively developing a research project.

The programme provides:

• Approximately 10 weeks of dedicated research time
• Experienced mentorship
• Research management support
• Access to a generous computing budget
• Interaction with other researchers and fellows
• Opportunities to engage with the University of Cambridge community
• Exposure to both technical AI safety and AI governance researchers

This environment can be particularly valuable for individuals seeking to transition into AI safety or governance research or deepen an existing research portfolio.

Professional Development and Long-Term Career Impact

ERA states that it invests deeply in both the professional development and long-term impact of its fellows.

Its support is built around individual research projects, high-touch mentorship and career support throughout the programme.

The goal extends beyond helping fellows complete a single research project. ERA aims to help participants develop into researchers, entrepreneurs and policymakers who can contribute to shaping the future of advanced artificial intelligence.

The fellowship can therefore provide participants with opportunities to:

• Develop specialised research expertise
• Build a substantive research portfolio
• Receive feedback from experienced researchers
• Explore career paths in AI safety and governance
• Connect with research organisations
• Develop professional relationships across academia, government and industry
• Become part of ERA’s broader alumni and professional network

Join a Global Network of ERA Alumni

ERA reports having a network of more than 200 alumni.

Its alumni have gone on to work at or become associated with a wide range of influential organisations spanning government, academia, AI research, public policy and nonprofit research.

Organisations listed by ERA as workplaces of its alumni include:

• UK AI Security Institute
• European AI Office
• Center for AI Standards and Innovation
• RAND Corporation
• Centre for the Governance of AI
• Centre for AI Safety
• Future of Life Institute
• Machine Intelligence Research Institute
• Institute for AI Policy & Strategy
• FAR AI
• Institute for Law & AI
• Center for Strategic and International Studies
• METR
• Apollo Research
• Geodesic Research
• Constellation Institute
• Safe AI Forum
• Anthropic
• Google DeepMind
• Microsoft
• SaferAI
• MIT
• University of Cambridge
• SecureBio
• Amodo Design
• Stanford University

The breadth of this network demonstrates the range of professional directions participants may explore following the fellowship.

Participation in the programme does not, of course, guarantee employment with any of these organisations.

More Than 30 Events During the Fellowship

The ERA Fellowship is not limited to individual research.

ERA states that it typically hosts more than 30 events throughout the fellowship period in Cambridge.

These events provide additional opportunities for participants to engage with researchers, experts, professionals and other members of the AI safety community.

The combination of independent research, structured mentorship and community events is designed to create an environment in which fellows can rapidly develop both their research and professional networks.

Build Lasting Connections in Cambridge

The fellowship also places considerable emphasis on community.

Over approximately two and a half months, participants work alongside other fellows, full-time researchers and members of Cambridge’s wider AI safety community.

ERA has a dedicated Community Health Manager who organises weekly community health activities intended to strengthen social connections between fellows and the wider Cambridge community.

ERA notes that many former participants remain connected years after completing their fellowships.

For applicants seeking not only research experience but also sustained relationships with others working on advanced AI safety and governance, this community component may be an important feature of the programme.

ERA Fellowship Winter 2027 Financial Support and Benefits

The fellowship provides substantial financial and practical support intended to enable selected participants to spend the full programme working on their research in Cambridge.

ERA states that fellows receive a competitive or generous stipend.

In addition, support includes:

• Fellowship stipend
• Lodging for the duration of the fellowship
• Transportation support
• Visa expense support
• Meals during working hours
• Research management support
• High-touch mentorship
• Access to a generous compute budget
• Career support
• Access to ERA’s wider professional and institutional network
• Participation in programme events and community activities

This support can significantly reduce the financial and logistical barriers associated with relocating to Cambridge for the 10-week programme.

International Applicants and Visa Sponsorship

The ERA Fellowship welcomes talented applicants from around the world.

ERA can provide visa sponsorship and visa-related expense support for fellows who require permission to travel to and participate in the programme in the United Kingdom.

However, international applicants should pay particular attention to the programme’s passport requirements.

Applicants requiring visa sponsorship must ensure that their passport:

• Has at least one blank page available for the visa vignette; and
• Is valid until at least 26 September 2027.

Anyone whose passport does not meet these requirements is advised to begin the renewal process as soon as possible because passport renewal can take time.

Who Is Encouraged to Apply?

ERA states that it welcomes talented individuals from around the world and at any career stage who are motivated to contribute to AI safety and governance research.

Applicants should not automatically exclude themselves because they believe they lack the conventional background for an AI fellowship.

ERA explicitly encourages people to apply even when they are uncertain whether they satisfy every criterion or feel they do not have the “right” background.

The programme is particularly interested in hearing from:

• Early-career researchers
• PhD students
• Working professionals
• People with backgrounds in computer science
• Engineers
• Legal professionals and researchers
• Individuals from related disciplines
• People interested in AI safety and governance who may come from other relevant backgrounds

ERA makes clear that this list is non-exhaustive.

Therefore, applicants from other academic or professional backgrounds who believe they can meaningfully contribute to the fellowship’s mission are encouraged to submit an application.

APPLY NOW

Do You Need to Be an Established AI Researcher?

Not necessarily.

One particularly important aspect of the Winter 2027 application guidance is ERA’s encouragement for people to apply even if they are uncertain whether they meet every criterion.

The organisation is specifically interested in early-career researchers, PhD students and working professionals.

That means applicants should evaluate their potential contribution based not only on whether they already have an established AI research career, but also on their relevant skills, research potential, professional experience, intellectual interests and motivation to address risks from advanced AI.

Applicants should nevertheless carefully review ERA’s research priorities before applying so that they understand the kinds of projects the programme seeks to support.

Read ERA’s Research Before Applying

Prospective applicants are strongly advised to familiarise themselves with ERA’s work before completing the application.

ERA particularly recommends reviewing its:

• Website
• Past Research
• AI Risk materials
• General applicant guidance from previous application cycles

Doing this should help applicants understand the types of research questions and approaches that align with the fellowship’s mission.

Candidates should also carefully consider the differences between AI Governance, Technical AI Governance and Technical AI Safety before choosing a track.

ERA Fellowship Winter 2027 Application Process

The initial stage of the application process primarily consists of short essay questions.

ERA estimates that this stage should take approximately two hours to complete.

Applicants are encouraged to prioritise the substance of their answers rather than spending excessive time polishing the writing.

Applications will be assessed on a rolling basis.

Applicants who successfully progress beyond the initial stage will be invited to complete two interviews.

The general process therefore consists of:

1. Review ERA’s website, research and AI risk materials.
2. Consider which of the three research tracks best matches your interests.
3. Complete the online application form.
4. Answer the required short essay questions.
5. Submit the application before the deadline.
6. Applications are reviewed on a rolling basis.
7. Candidates who progress will be invited to two interviews.
8. Successful candidates will proceed toward participation in the Winter 2027 fellowship.

Because applications are reviewed on a rolling basis, prospective applicants may wish to prepare and submit their application without unnecessarily waiting until the final deadline.

Important Warning About AI Use in the Application

Applicants should pay very close attention to ERA’s academic integrity policy.

ERA explicitly prohibits the use of AI tools to generate or refine written application responses.

According to the application guidance, all written responses must be entirely the applicant’s own work and should reflect their personal experiences, reasoning and insights.

External assistance is prohibited, including:

• Using AI tools to generate answers
• Using AI tools to refine application responses
• Having other individuals help produce responses
• Other forms of external assistance with written answers

ERA warns that applicants may be disqualified if it determines that their responses are not exclusively their own work.

Applicants should therefore complete all application essay questions independently and comply fully with ERA’s academic integrity requirements.

Important Application Form Information

Applicants should also be aware of two practical details when completing the form.

First, some application questions only appear after selecting a research track. Applicants should therefore make sure they have selected their intended track and completed all questions associated with it before submitting.

Second, the application form does not send confirmation emails after submission.

Applicants should not assume that the absence of a confirmation email means their submission was unsuccessful.

ERA Fellowship Winter 2027 Application Deadline

The deadline for the Winter 2027 ERA Fellowship is:

13 September 2026 at 11:59 PM Anywhere on Earth (AoE).

Applications submitted after the deadline may not be considered, so prospective candidates should plan accordingly.

The programme itself will run from 18 January to 26 March 2027 in Cambridge, United Kingdom.

Q&A Session for Prospective Applicants

ERA is also offering a Q&A session for individuals who have questions about the fellowship or application process.

Prospective candidates can register their interest through the registration option provided by ERA.

Before contacting the organisation directly with questions, applicants are encouraged to read the FAQ on ERA’s website.

If their question remains unanswered after reviewing the FAQ, the fellowship team can be contacted at hello@erafellowship.org.

Other ERA Programmes

ERA has developed a broader portfolio of programmes addressing risks from frontier technologies.

These include:

ERA

An intensive fellowship supporting researchers working on technical AI safety and AI governance.

The next cycle is Winter 2027, with applications currently open.

AI×Bio

A programme bringing together machine learning, biology, public health and policy to reduce extreme biological risks.

ERA has also indicated that an additional AI×Biosecurity Fellowship is planned for the same 18 January to 26 March 2027 period, with applications to be launched separately.

Frontier AI Security Residency

A project-led residency focused on areas including cybersecurity, hardware security, model protection and technical verification.

Technical AI Governance Forum

A programme convening technical and policy leaders to develop technical approaches and roadmaps for governing frontier AI.

Together, these programmes reflect ERA’s broader focus on developing talent, research and institutions capable of responding to challenges associated with transformative technologies.

Why the ERA Fellowship Winter 2027 Could Be an Important Opportunity

The ERA Fellowship offers an unusual combination of dedicated research time, mentorship, financial support, interdisciplinary collaboration and access to an established AI safety and governance community.

Selected fellows are given approximately 10 weeks to focus intensively on a substantive research project while receiving support from experienced researchers.

Several aspects make the programme particularly notable:

• It accepts applicants from around the world.
• Applicants can be at different career stages.
• ERA explicitly encourages candidates to apply even if they are uncertain about meeting every criterion.
• It supports technical and policy-oriented research.
• It provides a stipend and significant practical support.
• Visa-related costs and travel support are available.
• Lodging is covered for the programme duration.
• Fellows receive mentorship and research management support.
• Technical researchers can access a generous compute budget.
• Fellows join a community connected to major organisations across AI research, policy, academia and government.
• Participants can build a substantive research portfolio over an intensive 10-week period.

For candidates who want to build a career addressing risks from increasingly capable artificial intelligence, the fellowship provides an opportunity to devote substantial time to research while developing professional connections in Cambridge.

Key Dates and Details at a Glance

Programme: ERA Fellowship Winter 2027
Location: Cambridge, United Kingdom
Duration: 10 weeks
Programme dates: 18 January–26 March 2027
Application status: Open
Application deadline: 13 September 2026, 11:59 PM Anywhere on Earth (AoE)
Research tracks: AI Governance, Technical AI Governance and Technical AI Safety
Financial support: Competitive/generous stipend
Accommodation: Covered for the fellowship duration
Travel support: Available
Visa support: Available
Meals: Provided during working hours
Selection process: Initial application followed by two interviews for progressing candidates
Estimated initial application time: Approximately two hours
Application review: Rolling basis
Location requirement: Fellows participate in Cambridge, UK
International applications: Welcomed

Conclusion

The ERA Fellowship Winter 2027 in Cambridge, UK provides an intensive opportunity for researchers, PhD students, professionals and other talented individuals interested in tackling some of the most consequential challenges associated with increasingly capable artificial intelligence.

Running from 18 January to 26 March 2027, the 10-week programme combines full-time research with high-touch mentorship, research management support, career development, community engagement and access to ERA’s extensive network across academia, government, industry and the broader AI safety ecosystem.

With three tracks covering AI Governance, Technical AI Governance and Technical AI Safety, the programme provides multiple pathways for applicants with technical, policy, legal, engineering, computer science and interdisciplinary backgrounds.

ERA’s willingness to consider applicants from around the world and at different career stages also makes the programme accessible to candidates who may not yet consider themselves established AI safety researchers.

Successful fellows receive substantial practical support, including a stipend, lodging, transportation and visa expense support, meals during working hours and access to research resources such as a generous compute budget.

Prospective applicants should carefully review ERA’s website, past research and AI risk materials before submitting their applications. They should also remember that ERA strictly prohibits external assistance, including the use of AI tools, when writing or refining application responses.

Most importantly, interested candidates should take note of the application deadline: 13 September 2026 at 11:59 PM Anywhere on Earth (AoE).

Applicants who believe they could contribute to research aimed at understanding and mitigating catastrophic risks from frontier AI are encouraged by ERA to apply even if they are uncertain whether they meet every conventional criterion.

Visit the OFY website for more Fellowship Opportunities.