Platform review
DataAnnotation
Fast entry, real hourly rates, and a steady stream of worker reports of accounts closed without notice.
DataAnnotation is the quickest way into AI training work that pays by the hour rather than by the cent: no interview, no resume screen, no portfolio, just an unpaid assessment and an identity check. Workers who get in and get projects say the money lands on time through PayPal and that they set their own hours with no minimum commitment. The risk sits on the other side of that. Across Trustpilot reviews and BBB complaint records, workers describe losing access to the entire platform at once, with no reason given and no appeal, sometimes with earnings still pending. Treat it as supplementary income you could lose access to at short notice, not as a job.
Key facts about DataAnnotation
- Pay
- $25–$150+/hr advertised, by role
- Application
- Interview / vetted (per platform)
- Availability
- Not published by the platform
- Specialties
Read this first
The things that change how you should read everything else on this page.
- Many workers report being removed from the platform without notice. Across Trustpilot reviews and BBB complaints, people describe losing access to everything at once rather than to a single project, with no specific reason, no warning stage, and no route to contest it. Reviewers say the notice they got told them only that the account was “unavailable for new projects due to work below our customers' quality bar” (Trustpilot, July 2026). Wikipedia's Surge AI entry records “unexplained annotator account cancellations” as a criticism of the DataAnnotation platform, citing 2023-2025 press coverage. That is what workers and complaint records describe. No one publishes a figure for how often it happens, and DataAnnotation has not addressed it: it has not replied to negative Trustpilot reviews since claiming its profile in August 2023, and its own blog post about its reviews discusses assessment difficulty and work availability without mentioning terminations at all.
- Removal can strand money you have already earned. BBB complaints against DataAnnotation describe pending balances frozen on suspended accounts, including $5,576.25 in pending earnings and a single weekly earning of $2,889.86, and one complainant describes $400-$600 of promised document compensation never arriving. Trustpilot reviewers describe the same pattern at smaller amounts. None of the BBB records we read shows how any of these ended.
- The stronger claim, that DataAnnotation deactivates experienced high earners on purpose so it can replace them with cheaper newcomers, is an argument rather than an established fact. It is made at length in a January 1, 2026 Icy Tales opinion piece by Joshita, which asserts that “churn is not a side effect; it is the strategy” and rests on Reddit and Glassdoor testimony, an anonymous post on Blind, and pay-tier figures it does not source. Read the removal reports as removal reports. The motive behind them is not established by anything we could source.
- BBB's file is 12 complaints in three years. That is a detail source, not a rate, and the F grade is not a verdict on the work: BBB recorded 11 of the 12 as unpursuable because it could not locate the business, and 1 as unanswered.
- DataAnnotation's published rates do not agree with each other across its homepage, its FAQ, and its blog, and the gap reaches roughly a third at the specialist ceiling. The blog post also contradicts itself inside a single page, putting STEM work at $50-$100+/hr and coding at up to $50-75/hr. Any rate quoted for this platform, here or anywhere, only means something with the page it came from attached.
- In May 2025 a proposed class action was filed against Surge AI, DataAnnotation's reported parent, alleging it deliberately misclassified data annotators as independent contractors, with claims of unpaid training and near-impossible time limits that cut earnings. The Los Angeles Times report by Suhauna Hussain does not name DataAnnotation and does not state the court. Surge AI did not respond to the paper's request for comment. This is not a lawsuit against DataAnnotation.
- Applying means handing a government-issued photo ID and a face scan to a third party. Verification runs through Persona, and DataAnnotation's FAQ states that “Your information, including your biometric data, is stored by DataAnnotation and Persona” for identity verification. The homepage does not mention this step anywhere.
- DataAnnotation's blog quotes outside ratings for itself: 3.9 out of 5 from 300+ reviews on Glassdoor, and 4.1 out of 5 from over 1,500 reviews on Indeed. Those are the company's own citations of other sites, they carry no date, and the 3.9 is not the figure Glassdoor lists as DataAnnotation's overall score.
What the work is
DataAnnotation runs a marketplace where contractors review and grade AI-generated text. The company's own description of the work is “reviewing responses, checking accuracy, refining prompts, and rating outputs,” with other projects covering fact-checking, ranking model answers, labeling data, and subject-specific review. Workers sign on as independent contractors and the site describes no employment relationship. DataAnnotation is reported to be owned by Surge Labs Inc., the San Francisco company trading as Surge AI: The Verge and New York Magazine reported in 2023 that Surge AI appeared to own several separate work platforms, including Get Hybrid, Task Up, and DataAnnotation. DataAnnotation's own site names no parent company on any page we read.
Who they want
The company asks for strong reasoning, attention to detail, and subject-matter expertise, and says no coding or technical background is needed for general work. It runs separate tracks for coding, STEM, and professional fields including law, finance, and medicine, plus language-specialist work across roughly 40 languages. The gates it publishes are a Starter Assessment and a third-party identity verification requiring a government-issued photo ID and a selfie; there is no interview, no resume screen, and no portfolio review anywhere on the site. DataAnnotation publishes no country eligibility, on its homepage or in its FAQ, which is a real gap for a platform advertising work in dozens of languages and paying through PayPal.
Step 1
Sign up
The form at app.dataannotation.tech. No interview, no resume screen, no portfolio.
Step 2
Persona identity check
A government photo ID and a selfie; the FAQ says biometric data is stored.
Step 3
Starter Assessment
Unpaid, about an hour, one to two for specialized tracks, and there are no retakes.
Step 4
Approval
Typically a few days, though the company says it cannot reply to every applicant.
Step 5
Pick your own projects
Reviewers describe further unpaid qualification work before a paid task appears.
How you get in
You apply through the sign-up form at app.dataannotation.tech, reached from the Apply Now button on the homepage. Then you verify your identity. DataAnnotation uses a third-party provider, Persona, and requires a physical government-issued ID with a photograph of your face on it, such as a driver's license, passport, or national or state ID card, plus front-facing photos and turning your head side to side when prompted. The FAQ states that “Your information, including your biometric data, is stored by DataAnnotation and Persona” for identity verification. Next is the Starter Assessment. It is unpaid, the company puts it at about an hour, and specialized assessments in coding, math, chemistry, biology, physics, finance, law, medicine, or a specific language run one to two hours. You get one attempt: the FAQ says “There are no retakes or second chances.” The FAQ then says workers typically receive approval notification within a few days after completing the Starter Assessment, and warns that the company is unable to provide direct responses for every applicant. Once approved you pick projects yourself, and the company says high-quality performance opens more of them to you. Trustpilot reviewers describe further unpaid qualification work after approval before any paid task appears, one writing of “hours doing onboarding, getting qualifications etc.” (Trustpilot, July 2, 2026). The site describes no appeals process, for a failed assessment or for a closed account.
How it pays
No open roles to plot. We are not mirroring any open role on DataAnnotation's board right now, so there is no distribution to draw.
Pay is hourly on self-reported time, through PayPal. The homepage prices by role: $25-$50/hr generalist, $25-$40/hr language specialist, $60-$125+/hr for paralegal, nursing, and healthcare compliance, $75-$125+/hr for finance, accounting, legal, medical, and cybersecurity, and $75-$150+/hr for software and machine-learning engineering. The FAQ collapses the same work into broader tracks at lower numbers, $25-$30+/hr general, $20+/hr multilingual, and $50-$100+/hr for coding, STEM, and professional work, putting the FAQ's specialist ceiling about a third below the homepage's for comparable roles, while a company blog post gives a third set and contradicts itself inside one page, listing STEM at $50-$100+/hr and coding at up to $50-75/hr. Cadence depends on which page you read too: the FAQ describes deposits arriving a few days after you request them, the blog says PayPal or ACH on a weekly schedule. The FAQ says nothing about tax forms, 1099-NEC reporting, or self-employment tax, and the company publishes no guaranteed minimum and no minimum volume of work.
What workers say
The public record on DataAnnotation
Trustpilot's 1,961 reviews are polarized rather than mixed: 55 percent five-star, 20 percent one-star, and 4 percent in the middle. The two camps are rating different things. Five-star reviewers are rating money arriving and hours being free; one-star reviewers are rating being cut off from the platform, earnings they cannot withdraw, and support that does not answer, and some are applicants rating an unpaid assessment that never produced a decision. So the 3.9 average is the midpoint between two opposed experiences rather than a description of a middling one, and the removal complaint is consistent enough across Trustpilot and BBB records that Wikipedia's Surge AI entry carries “unexplained annotator account cancellations” as a documented criticism of the platform.
Scores, and what each one measures
- Trustpilot ↗3.9 out of 5
What it measures. Two opposed populations averaged together, not a middling experience. The five-star half is rating payment speed and schedule freedom. The one-star half is rating loss of platform access, balances they could not withdraw, and support that did not reply, and includes applicants rating an unpaid assessment that produced no decision. Only 4 percent of reviews sit in the middle, so the 3.9 headline describes almost nobody's actual experience.
- Better Business Bureau ↗F (BBB letter grade, not a star rating). Not BBB accredited.
What it measures. BBB's own complaint-handling record, not worker satisfaction and not job quality. The F grade is driven by responsiveness: BBB marked 11 of the 12 complaints unpursuable because it could not locate the business, and 1 unanswered. Twelve complaints across three years is far too small to read as a rate; what makes the file useful is the specificity of individual accounts, including named dollar amounts of stranded earnings.
Our read of it
Pros
- Payment landing on time is the thing workers say most consistently in DataAnnotation's favor. Trustpilot reviewers repeatedly describe PayPal payouts arriving promptly, and 55 percent of the profile's 1,961 reviews are five-star (September 12, 2026).
- Schedule flexibility is the next most common positive. Trustpilot reviewers describe setting their own hours with no shifts, no minimum commitment, and no scheduled availability.
- Entry is faster than on interview-based platforms: no interview, no resume screen, no portfolio stage. It is not free, though. The Starter Assessment is unpaid and runs about an hour by the company's estimate, you must clear a third-party identity check with a government ID and a selfie, and Trustpilot reviewers describe unpaid qualification work after approval before any paid task appears.
- Rates are published by role before you apply, which is unusual in this market. The homepage puts an hourly band against each named specialty, from $25-$50/hr generalist up to $75-$150+/hr for software and machine-learning engineering, so you can see what a track is advertised at before you spend an hour on the assessment.
- The work is reading, writing, and checking text. There is no customer contact, no phone work, no scheduled calls, and nothing to buy beyond a computer.
- The company states there are no upfront fees and that “We will never ask you for money or any other form of payment.” Its Trust and Safety page is given over to warning workers about impersonation scams run under its name, and names the only three recruiters it treats as legitimate, by first name and last initial. We found no worker reports of DataAnnotation itself charging fees.
Cons
- Workers report losing access to the whole platform without notice. On Trustpilot and in BBB complaints, people describe every project disappearing at once rather than a single assignment ending, with no specific reason, no warning stage, and no way to contest it. Wikipedia's Surge AI entry lists “unexplained annotator account cancellations” among criticisms of the DataAnnotation platform.
- Earnings already approved or pending can be frozen when an account closes. BBB complaints against DataAnnotation cite specific stranded balances, including $5,576.25 in pending earnings and a single weekly earning of $2,889.86. Trustpilot reviewers describe the same thing at smaller amounts, including one who says more than $1,600 was seized.
- Work droughts are routine and the company does not dispute them. Trustpilot reviewers describe weeks or months with no projects while their accounts stayed active and in good standing, one writing that four months in, the home screen still showed zero projects. The company's blog frames the pattern as inherent, saying “real AI development operates in cycles.”
- Support is widely described as not replying. BBB complainants report weeks of unanswered requests, and Trustpilot reviewers report silence on assessment status, account closures, and payment. Trustpilot's own profile disclosure says DataAnnotation has not replied to negative reviews, which it has been able to do since claiming the profile in August 2023.
- The unpaid cost of getting in is larger than the advertised hour. The Starter Assessment is unpaid, you get one attempt (“There are no retakes or second chances”) and no feedback, you must clear a government-ID and biometric identity check, and Trustpilot reviewers describe further unpaid qualification work after approval. Applicants also report completing the assessment and never receiving any decision at all.
- Quality scoring is opaque and there is no appeal. The FAQ confirms that access depends on past performance and warns that subpar or problematic contributions can cost you access to future projects, but it publishes no criteria, no thresholds, and no warning stage, and no appeals process appears anywhere on the site. Reviewers report receiving only a generic notice that their account is unavailable for new projects.
Who it fits, and who should skip it
Best for
- People who want to start this week without an interview
- Strong readers and writers who like close checking of text
- Specialists in law, finance, medicine, coding, or STEM wanting side income
- Anyone who needs unstructured hours with no minimum commitment
Who should skip it
Skip DataAnnotation if you need the income to be predictable. Work supply moves in cycles by the company's own account, your rate depends on which projects open to you, and workers report that access itself can end without warning, taking any pending balance with it. It is also the wrong platform if you would need someone to answer a support ticket, or if handing a government ID and biometric face data to a third-party verifier is not something you want to do.
Provenance
Where all of this came from
Every source behind this review
What each one says, and how much weight it deserves.
Advertised hourly rates attached to named roles: $25-$50 generalist, $25-$40 language specialist, $60-$125+ paralegal/nurse/healthcare compliance, $75-$125+ finance, legal, medical, accounting and cybersecurity, $75-$150+ software and machine-learning engineering. Also the company's description of the work and the Apply Now entry point.
How to read it. Marketing page. The rates are what the company advertises, not what anyone is shown to have earned, and they disagree with the company's own FAQ. The page names no parent company and states no country eligibility.
The Persona identity verification requirement, including government photo ID, selfie with head movement, and the statement that biometric data is stored by DataAnnotation and Persona. Starter Assessment length, its unpaid status and the single-attempt rule. PayPal payment with deposits a few days after request. Rate tiers of $25-$30+ general, $20+ multilingual, $50-$100+ coding, STEM and professional. The warning that subpar or problematic contributions can cost access to future projects.
How to read it. The company's own account of its process. It contains nothing about tax forms, 1099 reporting, country eligibility, or any appeals route, and its rate tiers sit well below the homepage's for comparable roles.
The company states it will never ask workers for money or any other form of payment, warns that people have used its name to commit fraud, and names three authorized recruiters by first name and last initial.
How to read it. Entirely about third-party impersonation. It says nothing about account removal, quality enforcement, or appeals.
The company acknowledges that project availability varies, framing it as cyclical because “real AI development operates in cycles,” and defends the one-attempt assessment. It cites its own Glassdoor and Indeed scores.
How to read it. The company answering its critics. It never mentions account terminations or workers losing access, which is the most common serious complaint against it, and its citations of other sites' scores carry no date.
Company claims of weekly PayPal or ACH payment, no upfront fees and no withheld percentages, over $20 million paid since launching in 2020, 100,000+ contractors worldwide, a 2.6% acceptance rate, and a tiered performance system.
How to read it. Unsourced company claims throughout. The 2.6% acceptance rate carries no method or date, the weekly cadence contradicts the FAQ's pay-on-request model, and the post puts STEM at $50-$100+/hr and coding at up to $50-75/hr on the same page.
3.9 out of 5 across 1,961 reviews, split 55% five-star and 20% one-star. Positive themes are prompt PayPal payment and free hours; negative themes are sudden loss of access, frozen balances, long droughts, and no reply from support. The profile was claimed in August 2023 and Trustpilot's own disclosure says the company has not replied to negative reviews.
How to read it. Self-selected reviewers, skewed toward people with a strong reason to post. The population is mixed: current workers, removed workers, and rejected applicants are all rating the same 3.9.
Reviews dated June and July 2026 describing accounts put on hold with no explanation, approved earnings that could not be withdrawn, four months with zero projects on the home screen, unpaid onboarding and qualification work before any paid task, and repeated unanswered support messages.
How to read it. Individual accounts, unverified by Trustpilot or by us. The generic suspension wording reviewers quote appears more than once, which makes it a template rather than an explanation.
F rating, not accredited, 12 complaints closed in three years and 8 in the last 12 months. Recurring themes of suspended accounts with pending earnings withheld, deactivation citing vague quality or terms-of-service grounds, and weeks without a support reply. Named stranded amounts include $5,576.25 pending and a $2,889.86 weekly earning.
How to read it. Twelve complaints is a very small file and cannot support a rate. BBB marked 11 unpursuable because it could not locate the business and 1 unanswered, so the F reflects unreachability as much as the complaints themselves.
Surge Labs Inc., trading as Surge AI, is a San Francisco data annotation company working with about 1 million annotators as of 2025. The Verge and New York Magazine reported in 2023 that it appeared to own Get Hybrid, Task Up, and Data Annotation. The Data Annotation platform has been criticized for lack of ownership transparency and for unexplained annotator account cancellations.
How to read it. A tertiary source summarizing press coverage. The ownership link is described as reported and apparent, not confirmed, and DataAnnotation does not acknowledge a parent company itself.
A proposed class action filed May 21, 2025 by plaintiff Dominique DonJuan Cavalier II, represented by the Clarkson firm, alleging Surge AI deliberately misclassified data annotators as independent contractors, with unpaid training and near-impossible time limits cutting pay. Surge AI did not respond to a request for comment.
How to read it. About Surge AI, not DataAnnotation, which the article never names. The article states no court. Allegations in a filed complaint are allegations.
An opinion piece arguing that DataAnnotation churns experienced workers deliberately because veterans are less profitable than slower newcomers, summarized in its own line that churn is the strategy rather than a side effect.
How to read it. Used only to show where the churn argument is made, never as evidence for it. It aggregates Reddit, Glassdoor and Blind posts, and its specific figures for unpaid qualification hours, worker lifetime earnings, and pay-tier deactivation risk are sourced to nothing checkable. None of its numbers appear on this page.
Evidence scorecard
What we can back up about DataAnnotation
11
0
Not one of them is work we've done ourselves. Read every line below as what somebody else states, discounted by its tier.
- Tier 1 · Verified first-hand0 accounts
We did it ourselves and can describe it, with a date. No dated first-hand account of DataAnnotation's own process is on record here.
- Tier 2 · Platform-published5 statements
Stated on the platform's own site, board, or in writing to us. Read off DataAnnotation's own pages, and cited to them.
- Tier 3 · Worker reports0 reports
Said by people who say they've done the work. We hold none for DataAnnotation.
- Tier 4 · Unverified claim6 claims
Printed by a marketing or aggregator page with no sourcing behind it. We repeat these labeled so you can recognize them elsewhere, never as fact.
Listed pay
Board not mirrored yetNo listings mirroredWe don't sync DataAnnotation's own board, so there's no listed rate here for us to count.
Effective pay
Not knownNobody has published itWhat a person clears per hour of their life after unpaid application time, unpaid training, idle time between tasks and self-employment tax. Nobody publishes it for DataAnnotation, and we can't tell you yet either.
Getting in
Interview / vetted (per platform)Tier 2 · Platform-publishedDataAnnotation's own material describes the step, and the review names the page it came from. What share of applicants clear it, DataAnnotation does not publish.
Specialty labels
Board not mirrored yetNo listings mirroredSpecialties are read off listings, and we don't sync DataAnnotation's.