Is data annotation legit? The real platforms, what they pay, and the scams to avoid
By Listicles research · Updated · Pay and listing counts checked on that date
Yes, data annotation is legitimate paid work, and DataAnnotation itself is a real company. It says it has paid contractors hundreds of millions of dollars since 2020, charges no sign-up fee and pays through PayPal, with general projects starting at $25 to $50 an hour. Outlier, micro1, Mercor and Alignerr are legitimate too. What is not legitimate is any "task" job that asks you to pay, deposit or buy crypto before you can withdraw your earnings.
The confusion is understandable. Real AI training platforms and fake "task" jobs look alike from the outside: remote, no interview at first, paid per task. The difference is which way the money moves. Real platforms pay you and never ask for money. Task scams pay you a little, then ask for a deposit. The FTC counted about 20,000 task-scam reports in the first half of 2024 alone, four times as many as in all of 2023.
The short version
DataAnnotation is legit. No fee, PayPal payouts, $25 to $50 an hour for general work and $40 to $150+ for coding, per its own FAQ.
The highest pay is on expert platforms. micro1 and Mercor both had a median top rate of $90 an hour across their open listings on 28 October 2026.
Outlier is real but quiet right now. It says it has paid experts $500M+, yet only 2 of its roles were in the public listings we track.
One rule spots almost every scam: a real platform never asks you to pay, deposit or "unlock" anything.
See every open role in one place on aitrainingjobs.xyz, which lists 1,729 jobs from 14 platforms.
Data annotation platforms compared
Pay is what each platform states, either on its own site or across its open listings on 28 October 2026. Every platform here pays the worker and charges nothing to join.
"Legit" is the question people are really asking, so it carries most of the weight. We ranked on five things. Whether the platform is open about pay before you work (30%). Whether it pays through a known, traceable channel and charges nothing (25%). How much work is actually open right now, measured from public listings rather than marketing claims (20%). The pay level for the typical role (15%). And how hard it is to get in, because a platform that rejects most applicants is not much use to you (10%).
For open work and pay we used live listing data from aitrainingjobs.xyz, which pulls every open role from 14 platforms' public feeds each day. For DataAnnotation, Outlier, Alignerr and Prolific, which aren't all in that feed, we used each platform's own pay statements, checked on 29 October 2026.
The 8 legitimate data annotation platforms, ranked
Legit, open about pay, and the easiest to start on your own schedule.
DataAnnotation is the company most people mean when they search "is data annotation legit", and the answer is yes. Its FAQ says it has "paid hundreds of millions of dollars to contractors since 2020", charges no sign-up fee and pays through PayPal, with money arriving "within a few days" of a withdrawal request. There is no minimum number of hours.
It also publishes its pay bands, which most platforms don't. General projects, such as rating chatbot answers and comparing AI outputs, start at $25 to $50 an hour. Multilingual work starts at $25 to $40. Coding starts at $40 to $150+, and its homepage advertises expert rates "up to $75–$125+/hr". On 29 October its site listed 112 open roles, 41 of them for languages.
The catch is the entry test. You take a written assessment, and if you don't pass you often hear nothing at all, which is where most "is it a scam?" posts come from. Silence after the test usually means you weren't accepted, not that the company has vanished.
Pay
$25–50/hr general, $25–40 multilingual, $40–150+ coding (its FAQ).
Payment
PayPal, on request. No fees.
Weak spot
No feedback if you fail the starter assessment. Higher tiers need a degree or years of experience.
micro1 had 252 open listings on 28 October 2026, more than any other platform we track, with a median top rate of $90 an hour. The range is wide, from about $30 to $300 an hour at the 10th and 90th percentiles, because it hires everyone from recent graduates to lawyers. 61 of its listings were marked beginner-friendly.
Screening runs through an AI recruiter, an automated interview that asks about your experience and tests skills live. It feels odd, but it is the real hiring step, not a scam. Most listings that name a country name the US, UK or Canada.
Pay
Median top rate $90/hr; range about $30–300/hr.
Open work
252 listings, 61 beginner-friendly (28 Oct 2026).
Weak spot
"Up to" rates: starting pay on a project is often lower than the top of the range.
Mercor matches experts to contracts for AI labs, screened by an AI interview. It had 221 open listings with a median top rate of $90 an hour, and its 10th percentile was $50, the highest floor of the large platforms. Science and engineering roles dominate, and the US, Germany and Spain are the countries its listings name most.
Only 36 of its listings were marked beginner-friendly, so Mercor is a poor first stop without a degree or professional track record.
Legit and well paid when there's work, but work comes and goes.
Outlier is Scale AI's contributor platform. Its homepage says it has paid experts "$500M+" and onboarded 700,000+ PhD and master's students, with a 4.1 out of 5 rating from 3,000+ experts on Trustpilot. It pays weekly, every Tuesday, by PayPal, AirTM or bank transfer, and shows you the rate before each project.
What it doesn't do is publish a standard rate, and its public pipeline was thin on our check: only 2 Outlier roles were in the feeds we track. Many people report long gaps between projects and sudden removals from a project. That is a reliability problem, not a legitimacy one.
Pay
Set per project, shown before you start.
Payment
Weekly on Tuesdays; PayPal, AirTM or ACH.
Weak spot
Unpredictable volume; few public listings right now.
Good rates for non-technical roles, backed by Labelbox.
Alignerr is run by Labelbox, a long-established data labelling company. Its site claims "$80/hr average pay" and 1,000+ people hired a week. Unlike most expert platforms, it advertises roles that need no technical background: an AI Data Quality Analyst role at $40 to $120 an hour "no technical experience required", and an English writing generalist role at $30 to $40.
The $80 average is Alignerr's own figure and we couldn't verify it from public listings, so treat the per-role ranges as the better guide.
Pay
$30–120/hr depending on role (its job pages).
Owner
Labelbox.
Weak spot
Headline average pay isn't independently checkable.
The legitimate route for students and recent grads.
Handshake, the campus careers network, runs AI training projects for its students, graduates and PhDs. It had 180 open listings with a median top rate of $80 an hour and a range of about $47 to $125. You get in by applying to a listing that matches your field and passing that project's assessment.
Pay
Median top rate $80/hr; range about $47–125/hr.
Open work
180 listings, 16 beginner-friendly.
Weak spot
Built around students and grads; others have fewer ways in.
Small but transparent: every listing states its pay.
Mindrift is Toloka's freelance AI tutor platform. All 19 of its open listings on our check stated an hourly rate, with a median top of $70 and a range of about $39 to $70. Work leans towards medicine, health and science. It is a sensible second platform, not a main source of hours.
Legit, low-stakes pocket money, but it isn't annotation work.
Prolific pays people to take part in research studies, including some that test AI. You can cash out to PayPal once you have $6 or £6, and it says you'll "always receive cash, never gift cards or points". It belongs on this list because people confuse it with annotation platforms. It's legitimate, but the studies are short and occasional, so it won't replace a contract on the platforms above.
The quickest way to see who is actually hiring today.
Checking a dozen platforms one by one is slow. aitrainingjobs.xyz pulls every open AI training role from 14 platforms' public feeds daily: 1,729 listings on 28 October 2026. Each one shows pay exactly as the platform states it and which countries can apply, and the site never lists a platform that charges applicants. It is free and needs no account.
How to tell a real data annotation job from a scam
The FTC's job-scam figures show how fast fake "task" jobs have grown. Reported losses to job scams rose from $90 million in 2020 to $286 million in 2023, then $223 million in just the first half of 2024. Task scams went from 5.6% of sampled job-scam reports in 2023 to 38.8% in the first half of 2024.
The money only ever moves towards you. If a "job" asks for a deposit, a fee, a training kit or crypto before you can withdraw, it's a scam. No platform on this list charges anything.
You found them; they didn't find you. Task scams usually start with an unexpected text or WhatsApp message. Real platforms recruit through their own sites and job listings.
The work is real work. Annotation means writing, rating and checking AI answers, often after a skills test. "Like these videos" or "boost these products" is the scam pattern the FTC describes.
Pay goes through a named channel. PayPal, bank transfer or AirTM, to an account in your name. Pay in crypto, or a balance shown only inside an app, is a warning sign.
The company exists outside the chat. Check that the site, the owner (Scale AI behind Outlier, Labelbox behind Alignerr, Toloka behind Mindrift) and the listing match. If you can't find the job on the company's own site, don't apply.
Which platform should you try first?
You want to start this week with no interview: DataAnnotation, then Alignerr.
You have a degree or professional expertise: micro1 and Mercor, where median top rates were $90 an hour.
You're a student or recent grad: Handshake AI.
You work in medicine or science: Mindrift, alongside Mercor.
You want to see everything that's open:aitrainingjobs.xyz, filtered by your field and country.
How we made this list
Scope
Platforms that pay individuals for data annotation or AI training work, open to US applicants, with no fee to join.
Criteria
Pay transparency (30%), traceable payment and no fees (25%), open work right now (20%), typical pay (15%), ease of entry (10%).
Sources
Each platform's own site and FAQ, checked on 29 October 2026; live listing data from aitrainingjobs.xyz (1,729 listings, 28 October 2026); FTC Data Spotlight on task scams.
Not tested
We did not complete paid projects on each platform for this list. Pay figures are as stated by the platforms or their listings; starting pay is often below the top of a range.
Updated
Questions
Is DataAnnotation legit?
Yes. DataAnnotation says it has paid contractors hundreds of millions of dollars since 2020, charges no sign-up fee and pays through PayPal within a few days of a withdrawal request. General projects start at $25 to $50 an hour and coding projects at $40 to $150+, according to its own FAQ checked on 29 October 2026.
How much does data annotation pay per hour?
Across 1,464 live AI training listings that stated pay on 28 October 2026, the median top of the hourly range was $75, with most between $25 and $145. Generalist work usually pays $25 to $50 an hour; coding, law, medicine and science roles commonly pay $80 to $150.
Why didn't DataAnnotation reply after my assessment?
DataAnnotation often sends no rejection message. If you hear nothing for several weeks after the starter assessment, you most likely weren't accepted. Silence is not evidence of a scam: the company pays through PayPal and charges no fees, which is the opposite of how task scams work.
How do I know if a data annotation job is a scam?
A real platform never asks you to pay, deposit money or buy crypto to start or to withdraw earnings. Task scams usually begin with an unexpected text or WhatsApp message and involve simple clicks with fake commissions. The FTC logged about 20,000 task-scam reports in the first half of 2024.
Which data annotation platform pays the most?
On 28 October 2026, micro1 and Mercor had the highest median top rate among the large platforms at $90 an hour, with Mercor's 10th percentile at $50. DataAnnotation's coding projects start at $40 to $150+ an hour. Highest pay goes to people with degrees or professional expertise in a specific field.
Sources
DataAnnotation homepage and FAQ (pay bands, payments, track record), 29 Oct 2026. dataannotation.tech/faq
aitrainingjobs.xyz platform comparison and listing statistics, data as of 28 Oct 2026, checked 29 Oct 2026. aitrainingjobs.xyz/platforms
Outlier homepage and FAQ (payments, track record), 29 Oct 2026. outlier.ai/faq
Alignerr homepage and role pages, 29 Oct 2026. alignerr.com