Get Paid to Train AI: 4 Real Programs, Direct Application Links + How to Pass the Screening
4 verified programs paying $30 to $3,850/week to train frontier AI. Direct application links + inside-view screening prep for Outlier, Mercor, Surge AI, and Anthropic Fellows. All open to international applicants. No PhD required.
Get Paid to Train AI: 4 Real Programs, Direct Application Links + How to Pass the Screening
You commented PAID on the carousel. Here's every application link + the inside view on what each program wants in your application.
All four are verified live in May 2026. All four accept international applicants. None require a PhD.
The pay ladder (quick reference)
| Tier | Program | Pay | Difficulty | Apply at |
|---|---|---|---|---|
| Entry | Outlier AI | $10-60/hr | Easy (signup + screening task) | outlier.ai |
| Specialist | Mercor | Up to $200/hr | Medium (apply + interview) | mercor.com/experts |
| Expert | Surge AI | $250-1,000+/hr | Hard (senior expertise required) | surgehq.ai/careers |
| Prestige | Anthropic Fellows | $3,850/wk + $15k/mo compute | Hard (research-style application) | alignment.anthropic.com |
1. Outlier AI — the easiest one to start this week
What you actually do: Rate AI model responses, spot hallucinations, write better prompts, sometimes write model-quality answers in your strong subject (math, coding, writing).
Pay reality check: Glassdoor + Indeed report $30-31/hr average, with task rates ranging $10-60/hr depending on project. Up to $75/hr during active tasking, but the queue is inconsistent — better as side income than primary.
How to apply:
- Go to outlier.ai and click "Sign up."
- Take their general assessment (writing quality + reasoning).
- If accepted, you'll see project listings. Each one has its own qualifier task. Pass the qualifier → unlock the work.
Screening prep:
- Their assessment looks for: clear writing, careful reasoning, ability to spot subtle errors. Treat it like a take-home test, not a quick form.
- The qualifier tasks for STEM projects (code review, math reasoning) pay higher. If you have those skills, prioritise those qualifiers.
- Don't rush the first task — your initial quality score affects what projects you get invited to.
Common mistake: Treating it like Mechanical Turk and rushing. Outlier rewards quality over quantity. Slow first week = better long-term project access.
2. Mercor — your white-collar skill pays you back
What you actually do: Use your professional skill (law, finance, medicine, engineering, journalism, accounting) to grade AI outputs in your field, write expert training data, evaluate frontier-lab reasoning.
Pay reality check: Up to $200/hr for in-demand specialists per Bloomberg (Apr 29, 2026). Lower for less specialised work. Weekly payouts via Stripe.
How to apply:
- Go to mercor.com/experts.
- Submit your CV + LinkedIn.
- Mercor's AI scans your profile. If matched to active projects, you'll get an interview invite (often async — record yourself answering).
- Pass the interview → get matched to projects.
Screening prep:
- Your CV is read by an LLM first. Use clear titles ("Senior Lawyer," "Practicing Physician," "Equity Research Analyst") not vague ones ("strategist").
- Highlight measurable expertise: years in field, deals closed, papers published, cases handled.
- For the async interview: be specific. "I worked on M&A at [firm] for 4 years, focusing on cross-border deals in tech" beats "I have legal experience."
Common mistake: Underselling your domain expertise. Mercor pays for specific expertise, not general skill.
3. Surge AI — top of the ceiling, top of the bar
What you actually do: Domain experts grade frontier AI on its hardest expert tasks. Medical fellows assess clinical reasoning, VC partners assess deal analysis, senior lawyers assess contract reasoning.
Pay reality check: $250-450/hr for medical fellows, $500-1,000/hr for VC partners on domain evaluations per their workforce page. Generalist rates start much lower ($20-40/hr) — the high rates are reserved for verified senior specialists.
How to apply:
- Go to surgehq.ai/careers.
- Apply to specific roles matching your senior expertise (browse "Workforce" listings).
- Submit credentials. They verify against your industry registry (medical board, bar, etc.).
- If verified, you get assigned high-rate projects.
Screening prep:
- Surge wants verifiable seniority. Submit registry numbers, license details, public publications.
- Apply to ONE role that matches your strongest credential, not multiple.
- Be patient — verification takes 1-3 weeks. They reject more than they accept.
Common mistake: Applying to multiple roles. Surge interprets this as "not actually a specialist in any of them." Pick one, lead with it.
4. Anthropic Fellows — the prestige play
What you actually do: 4 months full-time AI safety research at Anthropic with direct researcher mentorship. Work on alignment, mechanistic interpretability, AI control, model organisms, AI security, or model welfare. 80%+ of past fellows publish.
Pay reality check: $3,850/wk stipend + $15k/mo compute funding + relocation help. Total package is one of the highest in AI research training.
How to apply:
- Go to alignment.anthropic.com.
- Submit your application via the Anthropic careers portal (Greenhouse).
- The application asks for: a short research proposal (1-2 pages), CV, technical writeup of a project you've built or studied, and references.
- If shortlisted, you'll do a technical interview and a research discussion.
Screening prep:
- The research proposal is the highest-leverage part. They want: a concrete, testable question; awareness of prior work; a 4-month-scoped experiment. NOT a grand vision essay.
- For the technical writeup: pick a real project (an ML paper you reproduced, a system you built). Be specific about what worked, what didn't, and what you'd do differently.
- You don't need a PhD or published papers. Past fellows came from physics, math, cybersecurity, and self-taught backgrounds. They want signal, not credentials.
Common mistake: Vague proposals about "I want to work on AI safety." Anthropic gets thousands of those. A proposal that says "I want to investigate whether feature steering via SAEs generalises across model sizes, here's my 4-month plan" gets you to the interview.
Which one should YOU apply to first?
- No relevant expertise, want to start this week: Outlier AI. Sign up tonight.
- Have a degree + 2+ years of professional experience: Mercor. Your career so far is the qualifier.
- Senior in a specialised field (5+ years, verifiable credentials): Surge AI. The top of the ladder.
- Curious about ML research, willing to invest 4 months full-time: Anthropic Fellows. Apply to the next cohort.
You can apply to multiple — they're not exclusive. Many people start at Outlier for income while preparing a strong Mercor or Anthropic application.
What to do this week
- Today: Pick one tier. Sign up or start your application.
- Within 7 days: Submit. The bar to APPLY is lower than the bar to GET IN — most people never apply at all, which is the biggest mistake.
- Within 30 days: Have at least one active income stream (Outlier) OR one shortlist (Mercor / Surge / Anthropic).
If Anthropic is your goal: start your research proposal now, even if you have to revise it 10 times. Most rejected applicants didn't get rejected — they never finished the proposal.
Follow @africantechbro for more. Tag me when you get your first acceptance — I want to see the screenshot.