Q1 2027 AI Safety Fellowship
Call for mentors
A strong early-career researcher works full time on one of your research projects. You support them with two to four hours a week. We handle recruitment, screening, compute, logistics and day-to-day research management.
Advanced AI could go badly wrong for humanity. Pivotal exists to lower that risk. We find the strongest new researchers and put them to work on it, with people like you.
In Brief
Dates
18 January – 30 April 2027
15 weeks
Location
Remote mentorship is fully supported. Fellows will be in-person at the LISA coworking space in London.
Outputs
Research papers, policy reports, blog series, open-source tools. See examples.
Streams
AI governance, technical AI governance and AIxBio.
We welcome other projects that may still be a good fit! Please mention this in your application.
Compute
£3,000+/month per fellow
We want to support ambitious projects.
Cohort
> 30 fellows (~2% acceptance rate), typically 1–3 per mentor with group projects possible.
Strong graduate (primarily PhD) students, professionals, and outstanding undergraduates.
Who Does What
Each fellow has a Pivotal Research Manager. The work divides like this:
| Stage | You | Your fellow | Pivotal |
|---|---|---|---|
| Direction | Set the research direction | – | Help scope the project |
| Proposal | Approve the proposal | – | Draft, edit and format the proposal |
| Selection | Interview and choose your fellow | Applies, then does a work test and interviews | Source, screen and shortlist candidates |
| Research | Advise on research each week | Does the research, full time | Keep the research on track and unblocked |
| Networking | Help your fellow build connections | Joins the cohort at LISA | Provide networking and professional support |
| Authorship | Co-author, typically | Often first author | Co-author if warranted |
| Admin | – | – | Handle compute, stipends, visas, scheduling and admin |
Aims for the Fellowship
Aim 1: A high-quality output
Every project produces a high-quality output within the 15 weeks. This can be a paper, a policy report, a tool or something similar.
Aim 2: Career success
Every fellow ends the fellowship on track for an impactful career.
After the 15 weeks
Hired or funded
Fellows get hired, sometimes by their mentor's organisation, or receive follow-up funding to continue their research.
Extended by Pivotal
For a few highly promising projects that would not continue elsewhere, Pivotal provides extended funding.
Seed funding
Pivotal has the connections to secure seed funding for projects or organisations that grow out of the fellowship. If that looks promising, we're keen to help.
Timeline
- 1 Apply By Wed 16 September 2026
-
-
Step 1
Apply
A two-minute form, no commitment yet.
By Wed 16 September 2026 Now open -
Step 2
Short chat
Have a call with us about your project idea and motivation.
Mid-September
-
Step 1
Apply
- 2 Select fellows By Thursday 10 December 2026
-
-
Step 3
Send us the details
If accepted, provide your project description, logistics, and the information for your page on our website.
Late September -
Step 4
Choose your fellow
We provide shortlists, automated interviews and support for a work test. You run personal interviews and make the final decisions.
By Thursday 10 December 2026
-
Step 3
Send us the details
- 3 Q1 2027 Fellowship 18 January – 30 April 2027 (15 weeks)
Past Mentors
Our mentors have come from
Google DeepMind
UK AI Security Institute- RAND
IAPS
GovAI
Apollo Research
Redwood Research
Goodfire
Forethought
AI Futures Project
SecureBio
Timaeus
OpenAI
Mentors from our most recent cohorts
Peter Hase
Postdoc, Stanford University
Jide Alaga
Research Scientist, Google DeepMind
Cozmin Ududec
Head of Science of Evaluation, UK AI Security Institute
Adam Kaufman
Member of Technical Staff, Redwood Research
Kevin Wei
Adjunct, RAND
Emmie Hine
Researcher, Safe AI Forum
Samira Nedungadi
Head of Engineering AI, SecureBio
Patrick Levermore
Senior Policy Manager, Centre for Long-Term Resilience
What Mentors Say
-
Overall, the Pivotal fellowship is super well-organised, and mentors receive a great amount of administrative and organisational support in choosing and mentoring their fellows.
Ben BucknallPh.D. student, University of Oxford
-
Very smoothly organized with lots of support for the mentees and not too much overhead for me as a mentor.
Jasper GöttingDirector of AI, SecureBio
-
I was able to find high-quality people who understand AI safety to help me with a small project. They did amazing work and made me realize that this project is more important than I first realized. I'm glad Pivotal was able to support them while they did all of this.
Tyler TracyMember of Technical Staff, Redwood Research
-
Pivotal is my favourite program through which to mentor AI safety mentees. I've consistently had top applicants and found many amazing mentees through the program. The organizers respond quickly and have always been really helpful & accommodating!
Stefan HeimersheimResearch Associate, Adecco / Google DeepMind
-
The Pivotal mentees were highly talented and motivated. The program was well-organized and run. I would happily do this again.
Jesse HooglandExecutive Director, Timaeus
-
I was very impressed by the quality of applications I received via Pivotal and by the strength of the people I ended up working with.
Lewis HammondResearch Director, Cooperative AI Foundation
Where Our Alumni Are Now
Several past mentors have gone on to hire, co-author with or continue collaborating with their fellows.
- Kristina FortResearch Manager, LawAI. Previously Policy, Google DeepMind
- Soroush EbadianResearch Fellow, Anthropic
- Monika JotautaitėMember of Technical Staff, Apollo Research
- Elsie JangEmerging Scholar, Mercatus Center
- Theodore EhrenborgResearch Engineer, Apollo Research
- Nick MarshGrantmaker, Macroscopic. Previously Policy Researcher, AI Futures Project
- Hamza Tariq ChaudhryAI and National Security Lead, FLI
- Reilly HaskinsResearch Scientist, METR
- Patricia PaskovDirector of Standards, AVERI
- Aidan HomewoodTechnical Governance Lead, UK AI Security Institute. Previously Research Scholar, GovAI
Questions
A project a fellow can start in week one and get to a first result in ten weeks. See past projects for examples. Your Research Manager helps you scope the project before the fellowship starts.
Two to four hours a week for fifteen weeks. Before the fellowship, two or three scoping calls. During it, a weekly meeting with your fellow and availability on Slack. We also ask you to fill in a short survey before and after, and to help promote your stream. Selecting your fellow takes 2–12 hours in total, depending on how involved you want to be and how many applicants you interview.
By default, no. If this is a barrier for you, write to us and we'll do our best to support you.
Yes. Many mentors take two or three fellows, on separate projects or as a group on one larger project. We ask how many you want during onboarding.
Yes. We're happy to consider this and it has worked well in the past. Please indicate in the form if you're interested.
Yes. You and your fellow agree on a meeting time that suits you both. Your Research Manager covers day-to-day support in London hours.
Fellows do most of the research work. They are co-authors and often first authors. Mentors are typically co-authors. Pivotal never owns the IP. Where IP matters, fellow and mentor agree the arrangement case by case.