Deadline: 20th September 2026 | Join a fully funded 13-week research programme in London and work on technical research aimed at reducing risks from advanced AI.

London AI Safety Research (LASR) Labs is accepting applications for its Winter 2027 cohort, a full-time programme for technically capable people interested in moving into AI safety research. Participants work in small teams with experienced supervisors to develop a research project from initial proposal through experiments and publication.

Eligibility:

  • Open internationally; applicants do not need to be UK nationals.
  • Applicants without an existing right to work in the UK can receive visa support through the Government Authorised Exchange programme.
  • There is no specific degree or minimum professional-experience requirement.
  • Suitable applicants may range from undergraduates and PhD students to engineers and other experienced professionals.
  • Applicants should demonstrate strong technical and quantitative ability, particularly relevant machine-learning engineering skills.
  • Research ability, willingness to experiment under uncertainty and clear communication are important.
  • Previous experience building substantial projects, working with LLMs or independently developing relevant technical skills can strengthen an application.
  • The programme is full-time and primarily in person, so participants should be able to be based in London for most of the programme.

Description:

The Winter 2027 cohort runs from 11 January to 9 April 2027. The programme lasts 13 weeks and is based at the London Initiative for Safe AI in London, with participants expected to work approximately 40 hours per week.

During the first week, participants explore and critically assess several potential technical AI safety projects before being matched into teams of three to four. For the remaining 12 weeks, teams design and run experiments, refine their findings and work towards producing an AI safety research paper and accompanying blog post.

Research areas in previous cohorts have included AI control, untrusted monitoring, data poisoning, interpretability, chain-of-thought monitoring, sparse autoencoders, steganographic collusion and evaluations of potentially deceptive behaviour in advanced AI systems.

Participants receive regular supervision from experienced AI safety researchers. The programme also includes research talks, career activities, interaction with other researchers based at the London Initiative for Safe AI and visits to organisations working on advanced AI and AI safety.

Benefits:

  • £15,000 stipend to support living expenses in London.
  • Travel support.
  • Food during the programme.
  • Dedicated office and research workspace.
  • Supervision from experienced AI safety researchers.
  • Hands-on experience taking a technical research project from proposal to publication.
  • Opportunities to produce work suitable for submission to major machine-learning conferences and workshops.
  • Talks, career events and visits to organisations working in AI research and safety.
  • Access to a wider community of AI safety researchers and organisations in London.
  • Visa support for eligible international participants who do not already have UK work authorisation.

Find out more and apply at the official link below:

.
Apply before 20th September, 2026