A DIFFERENT ROUTE INTO AI SAFETY
Build something
worth backing.
Founder programmes, project grants and research residencies. See what each offers, what it expects and where the evidence stops.
What the JEV ratings mean
Two separate 0–3 assessments of the cited programme evidence. They do not measure your eligibility, acceptance odds or expected financial return. Funding amounts and deadlines are source facts, not model predictions.
- Founder support
- 0: not established · 1: precursor project support · 2: new-organisation funding · 3: structured launch support with funding access
- Technical safety
- 0: not established · 1: safety mission only · 2: explicit technical route · 3: primary hands-on technical safety activity
Fractional scores preserve JEV's position between rubric levels. Missing or stale assessments stay unrated. Source dates and unresolved terms still need checking before applying.
Liftoff Founders Week
LISA, London · 9–13 November 2026 · in person
Develop an AI-safety organisation with potential co-founders, a plan and budget, then discuss it with funders. No finished idea or co-founder required.
- Funding
- £1,000 participant stipend; travel and hotel covered. Project Tailwind funding is a separate decision, not an attendance award.
- Who it's for
- Founders, research leads and operators with evidence of building or relevant work. International applicants welcome; full-week attendance preferred.
Check before applying: Prepare evidence of work and a reason to found rather than join an existing organisation. Confirm the deadline's time zone with the organiser.
Sources · checked 2026-09-30
Catalyze Impact incubator
Location and attendance requirements not stated on the reviewed page
Personalised organisation-building support: direction, strategy, co-founder matching and mentorship. Supports both for-profit and nonprofit AI-safety organisations.
- Funding
- A founder stipend and access to a network of seed grantmakers and investors. Amount and investment terms not stated.
- Who it's for
- People considering founding an organisation in AI safety, security or resilience. Contact form available; a dated cohort intake is not confirmed.
Check before applying: Ask about stipend amount, selection process, geography and any equity terms before comparing offers.
Sources · checked 2026-09-30
Project Tailwind
International work considered; jurisdiction may add requirements
Coefficient Giving's founder call for new organisations addressing catastrophic AI risk. Selected expressions of interest lead to proposals or pitch discussions.
- Funding
- Pre-seed research or planning grants of $200,000–$2 million; larger later stages exist. A submitted expression of interest does not guarantee a response or funding.
- Who it's for
- Individuals or teams may seek pre-seed support without an entity. Best positioned for nonprofits; the FAQ says non-PBC for-profit initiatives are not expected to be funded.
Check before applying: Requires a direct catastrophic-risk case. The funder says AI-generated submissions are filtered. Write your own answers and clarify entity compatibility before relying on this route.
Sources · checked 2026-09-30
Transformative AI Fund
Confirm location-specific eligibility with the fund
Early-stage funding for individuals, new organisations and new projects within existing organisations. Replaces the closed Long-Term Future Fund.
- Funding
- Typical grants $10,000–$150,000, rarely above $300,000. Applications accepted continuously; project grants are not salaries or guaranteed awards.
- Who it's for
- Early-stage projects addressing transformative AI, primarily catastrophic-risk reduction. Supports technical research, demonstrations, training and field infrastructure.
Check before applying: Best considered for a bounded pilot with an explicit safety case. Check allowable costs and entity requirements in the application.
Sources · checked 2026-09-30
Sovereign AI · Trust, Safety & Assurance
UK venture fund focused on British AI companies
A commercial founder route for companies building AI evaluations, red-teaming, agent monitoring and audit infrastructure. The public request invites contact.
- Funding
- Venture investment, not a research grant. Investment size, valuation and dilution are not stated on the reviewed request.
- Who it's for
- Startups pursuing the fund's UK AI priorities. Incorporation and other investment eligibility need confirmation.
Check before applying: Adjacent to Liftoff, with different commercial and funding terms. A contact invitation does not mean an investment offer.
Sources · checked 2026-09-30
Meridian Visiting Researcher Programme
Cambridge · usually 3–12 months · at least three office days weekly
Workspace, research management and community for independent AI-safety and governance research. This is a residency, not a founder incubator.
- Funding
- No direct research funding. Workspace is free; accommodation and food require external funding.
- Who it's for
- Researchers able to work independently, including industry professionals. In-person participation required.
Check before applying: The page combines a February 1 spring-cohort banner with a rolling-applications FAQ. Current intake date needs confirmation; do not treat the banner as a new deadline.
Sources · checked 2026-09-30
ARENA 9.0
LISA, London · 5 October–6 November 2026 · in person
Hands-on technical AI-safety training. The current cohort is closed; the site offers an interest form for future iterations.
- Funding
- Travel, visa expenses, accommodation and programme-day meals covered. No cash stipend is stated.
- Who it's for
- Python ability, relevant mathematics and a genuine technical AI-safety direction. This develops skills rather than funding a company.
Check before applying: Future-cohort watchlist only. Do not apply to the closed ninth cohort expecting a live place.
Sources · checked 2026-09-30
BEFORE YOU APPLY
Bring evidence of what you can build.
Liftoff welcomes solo applicants without a finished idea. Start with a direction you care about and examples of work you have done.
Explore Project Tailwind directions ↗- Choose a safety problem. Explain the failure you want to prevent and why a new organisation would help.
- Select evidence. Link work you built or led. Separate demonstrated results from proposed research.
- Check practical terms. Confirm attendance, funding restrictions and whether the proposed entity is eligible.
- Write your own answers. Project Tailwind says it filters AI-generated submissions. Use this page for research, not as application copy.