Resource Guide · Updated May 2026
Get paid to train AI.
The honest guide.
AI labs spend over $1 billion a year on human feedback to train models like ChatGPT, Claude, and Gemini. Real people do that work, from home, on platforms that are actively hiring.
This guide lists every legitimate platform we know of. Honest pay ranges, real country restrictions, and no inflated numbers.
New platforms open signups regularly.
We will email you when something worth applying to launches. One email at a time, no spam.
What is AI training work?
The process is called RLHF — Reinforcement Learning from Human Feedback. When you chat with an AI, the quality of its responses traces back to thousands of people who rated, compared, corrected, and wrote examples for it to learn from.
Tasks range from comparing two AI responses and picking the better one, to writing complex prompts that expose model weaknesses, to producing high-quality content the AI trains on. The more expertise you bring, the more you can earn.
"The gap between a generalist ($15/hr) and a domain expert ($200+/hr) is entirely your existing knowledge. A doctor, lawyer, or senior engineer doing this work earns 10x what a beginner does."
One thing that applies across all platforms: you cannot use AI to complete the tasks. They detect it. The point of the job is your genuine human judgment, and bans are permanent.
Before you sign up
Work is project-based
Most platforms have dry spells between campaigns. Do not rely on a single platform for stable income.
Pay varies by country
The same task pays differently depending on where you are. US and UK workers typically earn more.
Assessments are real
You have to pass skill tests to get access to good tasks. Low-quality work gets you removed.
Advertised rates are ceilings
The $100/hr numbers are real but rare. Most beginners start at $15-25/hr and build from there.
Platforms open globally
These platforms accept applicants from most countries. No SSN or US residency required.
61 countries · Best for: Writers, coders, generalists
Realistic: Most people earn $20-35/hr. Top rates require STEM or coding expertise.
Most popular platform in this space. Requires a skill assessment to get in. Work arrives in projects and can be inconsistent between campaigns. Strong Reddit community for tips.
Global · Best for: Analytical thinkers, researchers
Realistic: Minimum $25/hr, including onboarding. Rates grow as your track record builds.
Quality-first platform. Tasks are cognition-heavy, not simple clicking. Builds a "Superhuman Profile" that unlocks higher-paying work over time. Apply at rex.zone.
AlignerrAI conversation, writing, annotation Most countries · Best for: Beginners, writers
Realistic: Generalist roles land at $14-20/hr. One of the easier platforms to get started on.
Short AI interview (10 min) to onboard. Good entry point for people new to this kind of work.
MindriftCreative writing, RLHF, storytelling Most countries · Best for: Writers, storytellers
Realistic: Creative tasks toward the lower end. Expert tasks and RLHF pay more.
Strong fit if you have a writing background. Tasks involve writing stories, dialogue, and narratives to train AI on human-quality creative content.
AppenSearch evaluation, annotation, transcription Global, 130+ countries · Best for: Beginners, multilingual speakers
Realistic: Most tasks pay $10-15/hr. Onboarding takes 2-6 weeks.
One of the oldest platforms in this space. Work is steady but pay is lower than newer platforms. Good for consistent part-time income.
Global · Best for: Multilingual speakers
Realistic: Rates vary by task type and language pair. Multilingual skills help a lot.
Apply via Aurora AI (their newer platform) for faster onboarding than the main Lionbridge site.
TolokaMicrotasks, image and text annotation Global, 100+ countries · Best for: Countries where other platforms are unavailable
Realistic: Honest average is $3-6/hr for standard tasks. Low pay.
Yandex-backed. Pay is genuinely low. Worth trying only if higher-paying platforms are not available in your region.
⚠ Low pay. Use as a last resort.
Global (payments limited in some regions) · Best for: Supplemental income
Realistic: Most workers earn below $10/hr in practice. Volume-dependent.
European-focused. Simple tasks, pay reflects that. Best used while building toward higher-paying platforms.
⚠ Low pay. Use as a stepping stone.
US and select countries only
These platforms restrict access by country. Higher pay on average, but you need to be in a supported region.
US, UK, Canada, Australia, NZ only
Realistic: Generalist roles start at $20/hr. Coding and reasoning tasks pay more.
Higher-quality tasks than most platforms. Assessment required. Expect 3-14 days to hear back after applying.
MercorAI evaluation, coding, domain expertise US, Canada, Australia focus
$20-400/hr
Intermediate to Advanced
Realistic: Floor is around $40/hr. PhD-level and medical/legal roles hit the top end.
Resume submission and AI interview to apply. One of the better-paying options for specialists. Worth applying even if you get turned down initially.
Scale AITechnical annotation, model evaluation US and select countries
Realistic: Competitive to get in. Technical roles command the higher end.
Works directly with major AI labs. Requires genuine technical expertise. Not beginner-friendly but pays well for the right background.
Realistic: Higher end for NLP and expert-level tasks. Harder to get accepted than most.
High quality work. Domain expertise in law, medicine, or STEM significantly improves your chances of getting accepted.
Open globally, but US workers get cash. Everyone else gets Amazon gift cards only.
Realistic: Research shows median is ~$2/hr after idle time. Only 4% of workers clear $7.25/hr.
The original crowdwork platform. Requires SSN for US cash payments. Non-US workers are paid in Amazon gift cards, not cash. Approach with realistic expectations.
⚠ Non-US workers get gift cards, not cash. SSN required for US cash payments.
⚠ SSN required for cash payments (US workers)
Find salaried roles on job boards
Beyond freelance platforms, companies post actual AI trainer roles with salaries, benefits, and stability. These are worth pursuing alongside platform work.
RemoteStack
Our own board. Search "AI trainer", "data annotator", or "RLHF" for verified remote roles.
Search →Handshake
Strong for students and recent grads in the US. AI labs post entry-level annotation and evaluation roles here.
Search →LinkedIn Jobs
Search "AI trainer remote" or "RLHF specialist". Set a job alert so new postings come to you.
Search →Indeed
Good volume of annotation and feedback roles. Pay is often listed upfront, which makes comparison easy.
Search →We Work Remotely
Smaller board, higher signal. AI training roles here tend to be from serious companies paying above average.
Search →How to spot a scam
This space has scams, especially on WhatsApp and Telegram. Legitimate platforms never do any of the following.
✗Ask you to pay to join, buy training materials, or purchase tools before starting
✗Promise guaranteed earnings or minimum monthly income
✗Ask for your SSN, bank details, or passport before you have been accepted and signed a contract
✗Have no verifiable company presence — no real website, no LinkedIn, no Crunchbase entry
✗Contact you out of nowhere via WhatsApp or Telegram offering high-paying AI work
Common questions
Is this real work or a scam?
The platforms listed here are legitimate. Outlier, DataAnnotation, Appen, and Scale AI are real companies you can verify. The work is real because AI labs genuinely need human feedback to train models. That said, pay varies widely and availability can be inconsistent.
Do I need technical skills?
Not always. Beginner platforms like Alignerr and Appen accept people with no technical background. But if you have coding, STEM, medical, or legal expertise, you can access much higher-paying roles. The gap is significant.
Can I work on multiple platforms at once?
Generally yes. Most treat you as an independent contractor and do not restrict you from working elsewhere. Check each platform's terms to confirm.
Can I use AI to do the tasks faster?
No. Every major platform detects AI-generated work using the same models you are evaluating. Bans are permanent. The entire point of your role is genuine human judgment.
I am outside the US. Where should I start?
Try Outlier (61 countries), Alignerr, Mindrift, Appen, and Lionbridge/Aurora AI. These accept international workers and pay in cash. Avoid MTurk if you want cash payments — non-US workers receive Amazon gift cards only.
Will this work disappear as AI improves?
Low-skill annotation tasks will likely shrink over time. But the highest-paying roles require domain expertise that is hard to automate. A doctor reviewing medical AI outputs is not going away soon. Higher-skill work is more durable.
Looking for a full-time remote job?
RemoteStack tracks 7,800+ verified remote roles across engineering, design, and product.
Browse jobs →