APPLICATIONS NOW OPEN · 22–28 NOV · 22–28 NOVEMBER · SUPPORTED BY KAIROS
Routing Generalists into AI safety.
A fully-funded week-long programme at an AI safety hub. You’ll work on a real problem with mentorship from a professional in the field, build connections, and leave with a clearer career strategy. The next cohort runs 22–28 November.
An in person week for people who make things happen. Take a problem from an AI safety organisation, ship something useful, and leave with a reference and connections.
Generalist paths into AI safety and where Gen Stream fits
There are several ways to move into AI safety as a generalist, from courses and mentored projects to residencies and full-time roles. This map shows some common routes and where Gen Stream sits among them.
Programmes select independently. Progression depends on fit and available opportunities.
04
Gen Stream · 22–28 November 2026
Build evidence. Open doors.
Spend five days in Oxford, fully funded, building a real AI safety project, with the next cohort running 22–28 November. We focus on what follows: work you can show, mentors who can speak to your ability, and introductions that help you move towards a role.
The cohort runs 22–28 November. From a rough idea on Monday to a demo you can share on Friday. Every day ends with something concrete in hand.
00/ 05
Before the week
↗BeforePreparation
Find your starting point.
Meet the cohort on an opening call, build shared context with a short reading list, and get matched with your mentor. Arrive at your accommodation the day before we begin.
1MondayScope & connect
Scope your project.
Meet the cohort at an AI safety hub, and spend the day narrowing your project to something you can finish within the week, alongside support from facilitators and time to speak with your mentor. You end the day by presenting your plan to the group.
2TuesdayBuild & experiment
Build a first version.
Two long work sessions, quick feedback rounds on your project, and a talk from someone working in AI safety. By the end of the day you should have a rough first version.
3WednesdayCritique & iterate
Review and adjust.
Draft your 30-day plan, then spend most of the day building to reach a working draft. There’s still time for small scope changes if your first approach isn’t quite working.
4ThursdayFinish & package
Prepare to share your work.
Finish your work, then record a short demo video and write a one-page summary explaining what you made and why it’s useful.
5FridayDemo & launch
Share your work and present it.
Send your work to people who can use it, book follow-up conversations, present at Demo Day, and finalise your 30-day plan. The week ends with a group dinner.
Meet some of our alumni, and the work they did with us.
The first Gen Stream cohort, together in London
Residentprojects.
Projects are real problems contributed by AI safety organisations and people working in the field. Each one is scoped so meaningful progress can be made towards it being used afterwards, and several have been continued or adopted since. Here are some examples.
Talent
Upskilling for rejected applicants
Project 01
CECaleb Eli PriceCohort 01 resident
Upskilling pathways for rejected fellowship applicants.
Field-building
AI safety fieldbuilding map
Project 02
GRGrace RobertsCohort 01 resident
A field map of AI safety fieldbuilding strategy.
Policy & education
US gov crisis response workshop
Project 03
NMNatasha MisraCohort 01 resident
A workshop on how the US government responded to past crises.
Evaluation
Impact evaluation for Gen Stream
Project 04
JXJian Xin LimCohort 01 resident
An impact evaluation and theory of change for Gen Stream.
Strategy
The case for future cohorts
Project 05
IBIgnacio BernardoCohort 01 resident
The case for future Gen Stream cohorts.
Data
AI safety salary dashboard
Project 06
ZTZuza TopolskaCohort 01 resident
A salary insights dashboard for AI safety roles.
Operations
Membership application process
Project 07
JSJessica SprockCohort 01 resident
A new membership application process for an AI safety hub.
Operations
New CRM for an AI safety hub
Project 08
SGShahil GoodkaCohort 01 resident
A new CRM for an AI safety hub.
Alumnivoices
CPCaleb PriceLisbon AI Safety Hub“
Everyone here is obviously very smart and agentic and has lots of ideas and connections, so to hear from them more in depth is great.
Gen Stream · Cohort 01
NBNacho BernardoBuilt a case for funding Gen Stream“
If I had not had such a specific space for generalists, for a single week, on this exact date, I'm 70% sure that I would have waited for the next Generator cohort to open.
Gen Stream · Cohort 01
DTDaniel Trevor WalshBuilt Glassdoor for AI Safety“
I got in touch with a lot of relevant people that I might have eventually gotten to know via a friend of a friend, in approximately what I assumed would be 3 to 6 months.
Gen Stream · Cohort 01
ZFZahra FarzanekhooBuilt aisafetyhackathons.com“
I now have a network and some perspective on the next steps that are actually in my reach, and how to go about them. I am also closer on what the best role for me is in AI safety.
Gen Stream · Cohort 01
ZTZuza TopolskaBuilt a salary insights dashboard for AI safety roles“
Building a tool for a grantmaking org would be very unlikely if not for this program, and it is directly relevant to my future plans.
Gen Stream · Cohort 01
Story 01 / of 05
Our team.
We design and run Gen Stream, and support residents beyond the week.
A fully-funded week at an AI safety hub. The next cohort runs 22–28 November — work on a real problem with a mentor from the field, build connections, and leave with a clearer career plan.
