Arnav Vijay · ftwsucks

Frontier AI & Interpretability

Curated research fellowships, bootcamps, and entry points for mechanistic interpretability and cognitive architecture.

Last verified October 24, 2026.

Timeline

Sunday, October 25, 2026

AI Collusion sprint submissions close (11:59 PM AoE).

Sunday, November 1, 2026

Pivotal applications close.

Friday, November 6

Swarm Alignment Hackathon. Runs through Sunday, November 8, 2026.

December 2026

SPAR Spring 2027 applications expected.

Monday, January 18, 2027

Pivotal fellowship begins.

Tuesday, January 19, 2027

MATS Winter 2027 begins.

Programs

Apart Research Sprints

Research hackathons

Open / Rolling

Weekend research hackathons where teams build an experiment, evaluation, dataset or tool.

Upcoming

Swarm Alignment Hackathon, November 6 to 8, 2026; AI x Epistemics Research Sprint, November 13 to 15; Agent Attribution Hackathon, November 27 to 29.

Details

Open to anyone, no prior affiliation needed; mostly online with some local sites. Top teams are invited to Apart Studio, and about 40% of Studio teams go on to the Apart Fellowship.

Next step

Sign up for a sprint and join the Discord.

ARENA

AI Alignment Research Engineer Accelerator

Closed

Fully funded in person ML bootcamp focused on ML engineering for alignment, including mechanistic interpretability and reinforcement learning.

Cohort 9.0

Cohort 9.0 runs in person at LISA in Shoreditch, London, October 5 to November 6, 2026; applications for 9.0 are closed.

Costs

Fully funded: travel to and from London, visa costs, accommodation, and meals on programme days.

Requirements

Genuine interest in AI safety, Python, and maths for modern AI (linear algebra, calculus, probability).

Process

Application form (under 90 minutes), coding test (six questions, one hour), 30-minute final interview.

Next step

Join the expression of interest form on the site for future cohorts. Future cohorts are announced on the site.

SPAR

Supervised Program for Alignment Research

Fall 2026 closedSpring 2027 expected

Part-time, remote, structured research programme with mentors, ending in Demo Day.

Details

3-month programme, 5 to 40 hours per week, ending in Demo Day. Spring 2027 applications expected around December 2026.

Eligibility

Open to undergraduates through professionals; no degree requirement; no prior research experience required.

Application

Answers must be written by the applicant without AI writing or polishing.

Next step

Use the Express Interest form on the site for the next round.

MATS

ML Alignment & Theory Scholars

Winter 2027 closed

Structured 12-week research programme across empirical and theoretical alignment tracks.

Winter 2027

Runs January 19 to April 10, 2027, 12 weeks, 40 hours per week, in person in Berkeley, CA and London, UK.

Stipend

$1,600 per week ($19,200 for the 12 weeks), plus travel, housing, office space, and weekday lunch and dinner.

Tracks

Empirical, Theory, Strategy & Forecasting, Policy & Governance, Systems Security, Founder & Field-Building, Biosecurity.

Eligibility

18 or older, any background.

Next step

Check the site for the next cycle. Do not use LLMs to write application text; MATS monitors for it.

LISA

London Initiative for Safe AI

Active

Shoreditch hub hosting AI safety programmes and organisations.

Next step

Use LISA as the hub for in-person AI safety programmes in London.

Pivotal Research Fellowship

AI Safety Fellowship

Closes Nov 1, 2026

15-week in-person research fellowship at LISA, London.

Dates

January 18 to April 30, 2027, in person at LISA, London.

Stipend

£15,000 to £18,000 (Senior Fellow £18,000), plus travel to London, £1,800 per month housing for fellows based outside London, and compute.

Support

Weekly 1-on-1 mentorship.

Process

Written application (1-2 hours, apply to at least one mentor), short video interview, mentor-specific work task, personal interview.

Next step

Apply before November 1, 2026.

AI Safety London

Community

Unconfirmed

Community meetups and reading groups in London.

Status

Not verified against a current listing; the community is listed on aisafety.com/communities.

Next step

Details unconfirmed; check the community site directly.

Sources

Details verified as of October 24, 2026. Dates and requirements may change; confirm on each programme's site before applying.