Applications are now open for the LASR Labs Winter 2027 cohort, a 13-week technical AI safety research programme designed for people interested in addressing some of the most important risks associated with advanced artificial intelligence.
The programme, run by London AI Safety Research (LASR) Labs, focuses on action-relevant research into concrete AI safety threat models, particularly the risks associated with advanced AI systems and potential loss of human control over increasingly capable models.
Selected participants will work full-time and in person in London, joining research teams of three to four people under the supervision of experienced AI safety researchers. Over the course of the programme, participants will develop an AI safety research project from an initial proposal through experimentation, analysis and publication.
The Winter 2027 cohort will run from 11 January to 9 April 2027, and successful participants will receive a £15,000 stipend to support their living expenses in London. LASR Labs will also provide food, office space and travel support during the programme.
Application deadline: 20 September 2026 at 23:59 GMT.
About the LASR Labs Winter 2027 Programme
London AI Safety Research (LASR) Labs is a technical research programme focused on reducing risks from advanced artificial intelligence. Its work concentrates on specific, action-relevant questions rather than AI safety as a broad or purely theoretical field.
The programme provides participants with an opportunity to gain hands-on experience conducting technical AI safety research while working alongside researchers and organisations within London’s AI safety ecosystem.
Participants are placed into teams of approximately three to four people and work under the guidance of an experienced supervisor. The objective is to take a research question from the early stages of project development through experimentation, iteration and eventual publication.
This makes the programme particularly valuable for applicants who want to transition into AI safety research, strengthen their technical research capabilities, build a publication record or explore a long-term career in AI safety.
Programme Dates and Duration
The Winter 2027 LASR Labs programme will take place over 13 weeks, from:
- Start date: 11 January 2027
- End date: 9 April 2027
- Duration: 13 weeks
- Format: Full-time and in person
- Location: London, United Kingdom
- Application deadline: 20 September 2026 at 23:59 GMT
The programme is run three times a year, giving participants multiple opportunities to engage with LASR Labs’ research environment.
What Does the LASR Labs Programme Offer?
One of the programme’s central strengths is its emphasis on taking participants through the entire research process rather than limiting them to theoretical learning.
The programme is divided broadly into an introductory Week 0 and a subsequent 12-week research period.
Week 0: Research Prioritisation and Project Selection
During the initial week, participants will be introduced to a selection of technical AI safety research projects.
With support from LASR Labs, participants will critically examine different research ideas and consider questions such as the following:
- What makes a research project promising?
- Which AI safety problems are particularly important?
- What approaches could generate useful evidence?
- How can researchers assess whether a research direction is likely to contribute meaningfully to AI safety?
- What different AI safety research agendas exist?
- How can researchers identify potential routes from research outputs to real-world impact?
Participants will then indicate their preferences regarding the available projects.
LASR Labs will use these preferences to form research teams and match participants with projects and supervisors.
This initial stage is designed to help participants develop research prioritisation skills, which LASR Labs considers important for producing useful and impactful AI safety research.
The 12-Week Research Project
After the introductory week, participants spend the remaining 12 weeks working with their teams to develop their selected research project.
The research process can involve:
- Developing and refining the research question.
- Formulating a research proposal.
- Designing experiments.
- Implementing technical approaches.
- Running experiments.
- Analysing results.
- Iterating based on findings.
- Developing conclusions and implications.
- Writing a research paper.
- Preparing the work for publication.
The expected output can take the form of a preprint, workshop paper or conference paper, accompanied by a blog post explaining the research and its significance.
Participants meet regularly with their supervisors throughout the programme. LASR staff also provide support with research planning and progress.
This structure gives participants practical experience with the full research lifecycle, from identifying and defining a problem to communicating findings to a wider technical audience.
£15,000 Stipend and Other Programme Benefits
A major benefit of the LASR Labs Winter 2027 programme is the financial and practical support provided to participants.
Successful participants will receive:
- £15,000 stipend to help cover living expenses in London.
- Food during the programme.
- Office space for conducting research.
- Travel support.
- Access to an in-person AI safety research environment.
- Supervision from experienced AI safety researchers.
- Opportunities to collaborate with other technical researchers.
- Exposure to leading AI safety organisations and researchers.
- Opportunities to develop and publish technical research.
- Opportunities to build professional relationships within the AI safety community.
The programme can therefore provide more than financial support; it can serve as an intensive entry point into the technical AI safety research ecosystem.
Research and Professional Networking Opportunities
Participants will work from the London Initiative for Safe AI, allowing them to interact with AI safety researchers and organisations on a day-to-day basis.
Throughout the programme, LASR Labs organises activities designed to expose participants to the wider AI safety field.
These include:
- Talks from leading AI safety researchers.
- A careers fair.
- Visits to AI research and safety organisations.
- Opportunities to interact with researchers and practitioners.
- Participation in events hosted within the London AI safety community.
Previous participants have reportedly used these interactions to find collaborators, mentors and additional career opportunities.
The programme can therefore be particularly useful for applicants who want to establish themselves within the AI safety research ecosystem and develop relationships that may continue beyond the 13-week programme.
AI Safety Research Areas
LASR Labs has previously supported research into several technical areas related to advanced AI safety.
Past projects have included work on:
- AI control and untrusted monitoring.
- Data poisoning.
- Interpretability probes.
- Chain-of-thought monitoring and faithfulness.
- Sparse autoencoders.
- Steganographic collusion.
- Scheming in frontier AI models.
- Evaluation awareness in advanced models.
The range of previous topics demonstrates that LASR’s research environment covers multiple approaches to understanding and mitigating risks associated with increasingly capable AI systems.
The programme has also worked with supervisors connected to organisations and academic institutions including Google DeepMind, the UK AI Security Institute, Anthropic and leading UK universities.
Strong Publication Record
LASR Labs may be particularly attractive to applicants who want to build experience in technical research and academic publication.
According to the programme, LASR research papers have been accepted at major machine learning venues and workshops, including NeurIPS, ICLR and ICML workshops.
Recent achievements include:
- Three LASR papers accepted to the NeurIPS 2025 main conference.
- One of those papers was selected for an oral presentation.
- A best paper award at an ICML 2025 workshop.
This publication environment provides participants with the opportunity to develop practical experience in producing research that can potentially reach established academic and technical venues.
Who Should Apply?
LASR Labs is looking for applicants who demonstrate a combination of technical ability, research ability and strong communication skills.
Importantly, the programme does not impose specific formal experience requirements.
Instead, applicants are assessed based on the skills, initiative and evidence they can demonstrate.
1. Technical Ability
Applicants should ideally have strong technical and quantitative capabilities.
Relevant experience may include:
- Machine learning engineering.
- Software engineering.
- Quantitative analysis.
- Mathematical or statistical work.
- Building technical systems.
- Conducting computational experiments.
- Working with large language models.
- Developing technical tools.
However, applicants do not necessarily need to have followed a conventional machine learning academic pathway.
2. Research Ability
Applicants should demonstrate an ability to investigate difficult problems and work effectively when the correct approach is not immediately obvious.
LASR particularly values people who can:
- Experiment systematically.
- Iterate when initial approaches do not work.
- Work independently under uncertainty.
- Identify useful research questions.
- Develop and test hypotheses.
- Analyse experimental results.
- Think critically about evidence.
- Connect technical research with potential real-world impact.
- Develop a theory of change for research projects.
Research ability is therefore not simply about having academic research experience. It can also be demonstrated through substantial independent projects, technical experimentation or other forms of problem-solving.
3. Communication Skills
Technical researchers must also be able to explain what their work means.
LASR, therefore, looks for applicants who can clearly communicate:
- What an experiment was designed to test.
- What the results demonstrate.
- What the results do not demonstrate.
- Why the findings matter.
- What implications the findings may have.
- How further research could build on the results.
The programme also highlights the importance of transparent reasoning, particularly when communicating technical research and its implications.
There Are No Specific Experience Requirements
One of the notable features of the LASR Labs opportunity is that applicants are not required to meet a narrowly defined list of academic or professional qualifications.
Instead, successful applicants are expected to have evidence that they can build, investigate, experiment and learn.
Potential evidence may include having:
- Built a substantial technical system.
- Produced a research result.
- Developed a useful tool.
- Conducted an analysis.
- Designed and completed an experiment.
- Developed a proof.
- Worked on a significant project in machine learning or another field.
- Worked with large language models.
- Undertaken serious independent technical projects.
- Taught themselves machine learning.
- Transitioned from another discipline into AI-related work.
- Started an independent research project.
This makes the programme potentially accessible to applicants coming from different academic and professional backgrounds.
Examples of Successful LASR Participants
LASR Labs highlights several examples of people who have previously been strong participants.
One example is an engineer with several years of industry experience who had built and deployed production systems. Although the individual had not initially worked full-time in AI safety research, they had been following the field and wanted to transition into research. They subsequently moved into work at the Australian AI Safety Institute.
Another example is a chemistry PhD student who had spent years independently reading and writing about AI safety. The LASR programme helped the participant turn that independent interest into published technical research. The individual subsequently moved to the Arcadia Alignment Project.
A third example is an undergraduate student who had not yet completed a bachelor’s degree but had established an AI safety group at their university and developed it into a substantial community. Following the LASR programme, the participant progressed towards an internship at Redwood Research.
These examples demonstrate that LASR does not exclusively seek applicants with conventional computer science or AI safety credentials.
Instead, applicants can demonstrate their potential through technical work, research, independent learning, initiative and evidence of commitment to the field.
Who Is Eligible to Apply?
The programme is intended for people who are interested in pursuing technical AI safety research and who can demonstrate relevant technical, quantitative, research or problem-solving ability.
Academic and Professional Background
Applicants may come from a wide range of backgrounds.
Potentially relevant profiles include:
- Machine learning engineers.
- Software engineers.
- Researchers.
- PhD students.
- Master’s students.
- Undergraduate students.
- Quantitative researchers.
- Scientists from other disciplines.
- Technical professionals transitioning into AI safety.
- Independent researchers.
- Self-taught machine learning practitioners.
- Individuals with substantial experience working with large language models.
A computer science or AI degree is not explicitly required by the programme information.
Likewise, previous AI safety employment is not presented as a mandatory requirement.
The strongest applications are likely to demonstrate evidence of technical capability, initiative and the ability to conduct meaningful research.
Nationality and Geographic Eligibility
The information provided by LASR Labs does not specify a restricted list of eligible nationalities or countries.
The programme is conducted in person in London, United Kingdom, so applicants should be prepared to participate full-time in London for the entire 13-week programme and should consider any applicable UK immigration, visa or work-authorisation requirements relevant to their circumstances.
Applicants from different countries and continents can therefore consider applying, provided they can satisfy the practical and legal requirements for participating in an in-person programme in the United Kingdom.
The programme’s published information also does not state that applicants must belong to a particular race or ethnicity. LASR instead encourages people from underrepresented backgrounds who are interested in the work not to rule themselves out because of doubts about their candidacy.
Why Applicants Should Consider LASR Labs
The Winter 2027 cohort could be particularly valuable for people seeking to move from general technical work into AI safety research.
Some of the major advantages include:
Intensive Research Experience
Participants spend 13 weeks working full-time on an actual AI safety research project rather than simply attending lectures or completing coursework.
Mentorship
Teams receive supervision from experienced AI safety researchers, providing participants with direct feedback throughout the research process.
Publication Potential
Participants work towards producing research that can potentially become a preprint, workshop paper or conference paper.
Strong Technical Environment
Working alongside AI safety researchers and organisations provides participants with exposure to current research practices and technical debates.
Career Development
The programme can help participants understand different career pathways within AI safety while providing opportunities to meet researchers, organisations and potential collaborators.
Financial Support
The £15,000 stipend is designed to support participants’ living expenses in London, while additional programme support includes food, office space and travel.
Research Prioritisation
Participants receive structured exposure to different technical AI safety research directions, helping them develop the ability to assess which research questions may be especially valuable.
What Kind of Candidate Is LASR Looking For?
A strong candidate does not necessarily need to have an impressive collection of academic credentials.
Instead, applicants should focus on demonstrating evidence of what they can actually do.
For example, a candidate could strengthen their application by showing:
- A substantial project they have personally built or completed.
- Technical competence through software, machine learning, quantitative or computational work.
- Research initiative demonstrated through an independent investigation.
- Experience with LLMs, whether professionally, academically or independently.
- Ability to learn independently, particularly where the applicant has taught themselves new technical concepts.
- Clear communication, including the ability to explain technical findings.
- Evidence of persistence, particularly when working on difficult problems without obvious solutions.
- An interest in AI safety supported by actual engagement rather than simply general enthusiasm.
- A focus on impact, including the ability to explain why a research project could matter.
Applicants should not assume that they are unqualified simply because they do not have a traditional AI safety background.
Application Deadline
The deadline for the LASR Labs Winter 2027 cohort is 20 September 2026 at 23:59 GMT.
Applicants should submit their applications before the stated deadline rather than waiting until the final hours, particularly because they may need time to prepare detailed information about their technical projects, research experience and motivation.
How to Apply
Interested applicants can submit their applications through the official LASR Labs application form.
Application deadline: 20 September 2026 at 23:59 GMT
Programme dates: 11 January to 9 April 2027
Programme location: London, United Kingdom
Programme duration: 13 weeks
Stipend: £15,000
Application: Apply for the LASR Labs Winter 2027 Cohort
Tips for Preparing a Strong Application
Although there are no rigid formal experience requirements, applicants should use their application to provide concrete evidence of their ability.
Highlight What You Have Built
Instead of simply listing skills, explain what you have actually created or accomplished.
This could be:
- A machine learning model.
- A software system.
- A research project.
- An analysis.
- A technical tool.
- An experiment.
- A substantial university project.
- An independent technical project.
Demonstrate Initiative
LASR specifically values people who have taken meaningful steps independently.
If you taught yourself machine learning, started an AI safety project, experimented with LLMs or pursued research outside your formal studies, explain what you did and what you learned.
Explain Your Research Thinking
Do not only describe what you worked on. Explain how you approached the problem.
For example, discuss:
- The question you were trying to answer.
- Why the question mattered.
- Your hypothesis.
- Your methodology.
- What you expected to find.
- What actually happened.
- How you responded to unexpected results.
- What you would do differently in a future experiment.
Show Your Ability to Learn
AI safety is a rapidly developing research field. Demonstrating that you can quickly learn unfamiliar concepts and apply them to difficult problems can be valuable.
Avoid Self-Rejection
LASR Labs specifically encourages interested applicants, particularly those from underrepresented groups, not to exclude themselves prematurely because they doubt the strength of their candidacy.
If you can demonstrate technical ability, research potential and genuine interest in the field, it may be worthwhile to apply even if your background does not perfectly match the examples of previous participants.
Visit HERE for information about the LASR Labs Winter 2027 AI Safety Research Programme.
Conclusion
The LASR Labs Winter 2027 cohort offers an intensive pathway into technical AI safety research for applicants who want to work on concrete questions surrounding the risks posed by advanced artificial intelligence.
Running from 11 January to 9 April 2027, the 13-week programme combines research training, hands-on experimentation, mentorship, professional networking and publication opportunities. Participants will work in small teams under experienced supervision and take a research project from proposal through experimentation and analysis to a potential publication.
The programme is also notable for its flexible approach to applicant backgrounds. LASR Labs does not require a specific degree, professional history or number of years of experience. Instead, it looks for evidence of technical ability, quantitative skills, research potential, initiative, experimentation and clear communication.
With a £15,000 stipend, additional food, office space and travel support, as well as access to London’s AI safety research ecosystem, the programme could be a significant opportunity for engineers, researchers, students and independent technical practitioners seeking to build a career in AI safety.
Interested applicants should prepare their applications early and submit them before the 20 September 2026 at 23:59 GMT deadline.
For more global Fellowship opportunities, visit OFY HERE.
