MARS 3.0

MARS (Mentorship for Alignment Research Students) connects aspiring researchers - from students to professionals - with experienced mentors to conduct AI safety research. Over 2-3 months, participants work on focused projects in either technical or policy domains, part-time, on a volunteer basis.


Participants in previous iterations have gone on to publish their work at venues such as NeurIPS, ICML, EMNLP & ControlConf, mentored by researchers at Redwood Research, Apollo Research, Google Deepmind, University of Cambridge & others.

Application deadline: 18th May, 23:59, Anywhere on Earth time.
(For both participants entering Stage I. and mentor applications.)

See more information for mentors here.

Program Details

MARS (Mentorship for Alignment Research Students) connects participants with experienced mentors to conduct AI safety research from July 7th to mid-late September 2025. Over this period, participants work on focused projects in either technical or policy domains. (Exact dates may be subject to minor adjustments.)

Why apply?

  • Work directly with experienced AI safety researchers on research projects

  • Join an intensive week in Cambridge (UK) with workshops, speakers, and accommodation provided

  • Receive regular mentorship and project support throughout the program

  • Opportunity to publish your work (previous participants have published at EMNLP and on arXiv)

  • Connect with a community of researchers interested in AI safety

Format & Time Commitment

MARS begins with an intensive week in Cambridge from July 7th-13th 2025, where participants come together to kick off their projects. We provide accommodation, meals, and travel support (~$100 for participants & mentors traveling from the UK, ~$500 for longer flights) during this period.

Following the in-person week, research continues remotely until September 2025. Teams meet weekly with their mentors and research managers, devoting approximately 8 hours per week to their projects. For teams based in Cambridge, Oxford, or London, we also organise optional in-person co-working sessions.

Throughout the program, CAISH provides hands-on support to help teams develop their research, culminating in final project presentations in September 2025.

Participation in the program is not paid, is understood to be undertaken on a volunteer basis as a learning opportunity.

Application Process

Stage I is due May 18th. The process begins with a general application. Selected candidates may be asked to complete a coding test or written assignment as part of this initial evaluation.

In Stage II, promising candidates are evaluated for specific research teams. We review applications on a rolling basis as both mentor and participant applications come in. During this stage, mentors may request additional materials to assess fit, such as targeted questions, interviews, or take-home assignments. We may also conduct brief interviews or work tests before advancing candidates to mentor-specific evaluations. Final decisions will be made by June 16th.

  • Time Commitment

    Participants should expect to each spend 8-15+ hours/week on their project, depending on the project chosen.

    Weekly Check-Ins

    Throughout the duration of the program, you are expected to check-in:

    • With your teammates once a week. It'll be up to your team to coordinate meeting times or create coworking systems.

    • With your supervisor once a week.

    The format of check-ins with your supervisor will be at their discretion.

    We will also have program-wide events and check-ins, including socials, co-working sessions, andguest speakers. More details on these events will be announced to accepted participants.

    MARS Research Symposium

    This event will take place in ~mid-late September, where each reseach group presents their research. Precise details to come


  • We welcome applications from students and professionals, particularly those with backgrounds in:

    • Computer Science, Mathematics, or related technical fields

    • Cybersecurity or Hardware Engineering

    • Policy, Philosophy, or Social Sciences (for governance track)

    • Other relevant fields with strong analytical skills

    For Technical Track:

    • Strong foundation in programming

    • Background in machine learning

    • Additional prerequisites vary by project

    For Governance Track:

    • Strong analytical and writing skills

    • Interest in AI policy

    • Specific requirements vary by project

    Note: We encourage applications even if you don't meet all prerequisites. The general application will be followed by project-specific evaluations. We value potential and motivation over previous research experience. Both recent graduates and promising undergraduates are welcome to apply.

    While the program is primarily designed for UK-based participants (especially around Cambridge, Oxford, and London), we welcome international applicants who can:

    • Attend the in-person week in Cambridge (July 7th-13th)

    • Commit to regular meetings in UK/European time zones

    • Dedicate ~8 hours per week to the project

  • Virtual mentorship is possible, but we may give preference to those who can come to the in-person kick-off week in Cambridge.

    Mentorship time

    Mentors are free to choose when and how they meet with mentees. We expect mentors to commit ~10 hours to the project in the kick-off week, and weekly or biweekly check-ins with teams afterwards.

    Project

    Mentees will work part-time on this project. They will vary in background and research experience. The ideal project should allow mentees to reach meaningful milestones in 2-3 months of part-time work.


For any additional questions, email gabor@cambridgeaisafety.org.