By Mufy Pachorawala, founder of getAIwork · Last updated: September 18, 2026
Key facts
- Outlier is Scale AI. Its terms, effective May 11, 2026, name “Scale AI, Inc., a Delaware corporation, and its affiliates, including Smart Ecosystem Inc., doing business as Outlier AI”, and cover Outlier, Remotasks and the Human Frontier Collective jointly
- DataAnnotation is Surge Labs, Inc., per its terms last updated July 14, 2026. That name appears nowhere on its marketing site, its privacy notice or its signup page
- Outlier publishes zero dollar figures anywhere on its public site. DataAnnotation publishes rate figures on four pages, and they do not agree with each other
- Outlier requires at least an associate degree, a CV and a LinkedIn profile. DataAnnotation asks for a name, email, password and phone number
- DataAnnotation gives you one attempt at the Starter Assessment, with no retakes. Outlier does not publish a retake policy at all
- 618 of 1,662 live listings on the free getAIwork board are paid platform programmes like these two, as of September 24, 2026
Outlier vs DataAnnotation: the comparison
| Outlier | DataAnnotation | |
|---|---|---|
| Operating entity | Scale AI, Inc. and affiliates including Smart Ecosystem Inc. d/b/a Outlier AI | Surge Labs, Inc., a Delaware corporation |
| Where that is disclosed | On the front page of every legal document and in the site footer | Only inside the click-through terms, reachable from the signup form |
| Terms last dated | Effective May 11, 2026 | Last updated July 14, 2026 |
| Signup asks for | CV, LinkedIn profile, valid ID, mobile phone, and at least an associate degree | First name, last name, email, password, phone |
| Country policy | An allowlist of roughly 100 permitted countries, published | No country list published anywhere. The vendor says only “available in 50+ countries” |
| Published rates | None. Zero dollar figures on the public site | A public rate card, from $25 to $150 or more an hour depending on track and which page you read |
| Payment processors | PayPal, Airtm or ACH | PayPal |
| Payment schedule | Weekly, processed Tuesdays for the prior Tuesday to Monday | On request. “Withdraw anytime”, delivered within a few days |
| Payout minimum | Not published | Not published |
| Onboarding | 30 to 90 minutes, then project-specific qualification | One Starter Assessment of about an hour, one to two hours for specialist tracks |
| Retakes | Not addressed anywhere | None. One attempt per track, stated twice |
| ID verification | Required. Vendor not named | Required. Persona, named by the vendor |
| Appeal against deactivation | None published | None published |
Compiled September 17, 2026 from each vendor’s own live pages, terms and FAQs. Pay figures are quoted only as the vendor lists them and are not earnings anyone is promised.
Who would you actually be working for?
This is the cleanest difference between them, and it has nothing to do with the work.
Outlier tells you immediately. The first sentence of its terms of use names Scale AI, Inc. and Smart Ecosystem Inc. doing business as Outlier AI, its privacy policy repeats it, its footer says “powered by Scale AI”, and it gives a postal address at 650 Townsend Street in San Francisco. The same terms also cover Remotasks and the Human Frontier Collective, so all three platforms run on one contract. Scale AI itself appointed a new chief executive, Francis deSouza, on July 30, 2026, with founder Alexandr Wang staying on as chairman.
DataAnnotation does not. Its operating company, Surge Labs, Inc., appears exactly once across the whole web property: inside the terms of use at app.dataannotation.tech/contract/280, which is reachable only from a link at the bottom of the signup form. The name is absent from the FAQ, the about page, the privacy notice and every footer. The privacy notice names no company at all, only “DataAnnotation.tech and each of its owners and affiliates”, and gives no postal address. You have to begin signing up to find out who you would be contracting with, and neither brand publicly acknowledges the connection between DataAnnotation and Surge AI in any other document.
Neither of these is a scam signal. Both are real companies. It is a difference in how much they expect you to take on trust, and it is worth noticing before you hand over a government ID.
Which is easier to get into?
DataAnnotation, by a wide margin, at the front door. Its signup form asks for five fields and no CV. Outlier wants a CV, a LinkedIn profile showing education and work history, a valid government ID, a mobile phone from your country of residence, and says plainly: “Yes, you need at least an associate degree to work on Outlier.”
The trade is at the next step. Outlier spreads its screening across a 30 to 90 minute general onboarding and then project-specific qualification tasks, and nowhere says you get only one go. DataAnnotation compresses everything into a single Starter Assessment and then closes the door: “You can only take the Starter Assessment once. There are no retakes or second chances.” It says that twice on the same page. We wrote a separate piece on how to approach that assessment, because one attempt changes how you should spend the evening.
| Country policy | What the vendor publishes |
|---|---|
| Outlier | An allowlist of roughly 100 countries in its flexible working guidelines, effective July 15, 2026, plus export-control exclusions covering Cuba, Iran, North Korea, Sudan, Syria and Ukraine. VPNs to mask location are explicitly banned |
| DataAnnotation | Nothing. The only geographic statement anywhere is “available in 50+ countries” on its about page, with no list. The signup form has no country field and notes only “please include the country code if outside the US” |
If you are outside the obvious markets, Outlier can tell you in thirty seconds whether you are wasting your time. DataAnnotation cannot, which is a real cost even though it looks like openness.
The getAIwork board is free and public, screened by a human. Or take the 2-minute match quiz to narrow it to what fits you.
Take the 2-minute match quiz →
What do they pay?
Outlier publishes nothing. We read the full text of its terms, FAQ, community guidelines, flexible working guidelines and privacy policy, and there is not a single dollar figure on the public site. Its position is that “rates vary by expertise, project complexity, and location” and that “you’ll always see the tasking rate before starting any project”, which moves the disclosure inside the logged-in product where nobody can check it. Payment is weekly, processed Tuesdays for the previous Tuesday to Monday period, via PayPal, Airtm or ACH, in US dollars. No minimum payout is stated.
DataAnnotation publishes a great deal, and it disagrees with itself. Four of its own pages quote different figures for the same tracks:
| Track | FAQ and about page | Homepage and role cards |
|---|---|---|
| Generalist | $25 to $30 or more an hour | $25 to $50 an hour |
| Coding | $50 to $100 or more an hour | $75 to $150 or more an hour |
| STEM | $50 to $100 or more an hour | $75 to $125 or more an hour |
| Professional (law, finance, medicine) | $50 to $100 or more an hour | $75 to $125 or more an hour |
| Multilingual | $20 or more an hour | $25 to $40 an hour |
All five rows differ between two sets of live vendor pages, read on the same day. The blog adds a third set again. Figures are as listed by the vendor and are not earnings anyone is promised.
The same page-to-page problem shows up in its headline statistics. Its FAQ says “we have paid over $20 million to contractors since 2020”, while its about page and every role page say “$150M+ paid to contributors”. That is a gap of seven and a half times between two live pages on the same website. Its about page also shows “120M+ annotations completed” and “13B+ annotations to date” in the same block, and displays the placeholder number 42 as the count for all nine of its domains.
Neither platform publishes a payout minimum. Both reserve the right to reject work and not pay for it: Outlier may withhold payment “in full or in part” on quality grounds, and DataAnnotation “reserves the right to reject any Work Product… that it deems inadequate for any reason in its discretion”, after which “you shall not be entitled to payment”.
Assessments, side by side
Outlier’s flow is three stated stages: platform onboarding and verification, project qualification with assessment tasks, then tasking, “and start earning”. The wording implies earning begins at stage three, which implies stages one and two are unpaid, but the vendor never says so outright. Nor does it publish a review time or a retake policy.
DataAnnotation’s flow is one gate. Choose a track at signup, take the Starter Assessment, get an email “within a few days”. Most take about an hour, specialist tracks one to two. The vendor calls it “free”, which means free of charge rather than paid, and it never states whether your time is compensated. It also explains why there are no retakes: multiple attempts would let applicants “pattern-match from feedback rather than demonstrating genuine capability”.
What happens if you get deactivated?
The same thing on both, which is to say very little in your favour. Outlier’s terms: the company “may terminate these Terms, deactivate or suspend your account… at any time, immediately without prior notice for any reason”, after which “you will not be able to retrieve any information related to your account”, though it does promise “a final payment for Tasks already completed”. DataAnnotation’s terms: “We reserve the right to terminate your Labeling Services… at any time and for any or no reason in our sole discretion, without liability to you or any third party.”
Neither document contains the word “appeal”, or any equivalent. Outlier has a channel for reporting other people’s conduct but none for contesting your own deactivation. DataAnnotation’s only route is an in-product support form, which is not much use to someone already locked out. In practice DataAnnotation’s version is quieter still: work simply stops appearing, which its FAQ describes as the consequence of “subpar or problematic contributions”, with no formal termination event at all. We cover what that looks like from the inside in Outlier account deactivated and DataAnnotation no response.
Both contracts also require individual arbitration and waive class actions, which means the real volume of disputes on either platform is structurally invisible to anyone outside.
What the public court records actually show
We searched US federal dockets on September 17, 2026. Two findings, and both cut against the way this is usually reported.
- The Scale AI misclassification case is over. Ramey v. Scale AI, Inc., filed in the Northern District of California in October 2024 under the Fair Labor Standards Act with HireArt as co-defendant, settled. A joint notice of settlement was filed in February 2025 and the case was dismissed on May 12, 2025. Any article describing it as ongoing is out of date.
- The other worker-side case is also closed. Schuster v. Scale AI, Inc., which named Scale AI, Smart Ecosystem and Outlier AI as defendants, was filed in January 2025, met a motion to compel arbitration that was never decided on the public docket, and ended in a voluntary dismissal on August 10, 2026. Terms are not public.
- DataAnnotation and Surge Labs have no federal litigation footprint at all. Searches for both names returned nothing.
- Neither company has ever filed a California data-breach notice. We checked the state registry with a control search to confirm it was working.
Those last two are absences of record rather than clearances, and the arbitration clauses mean neither company’s dispute volume can be measured from outside. We would rather be precise about that than imply either more or less than the records support.
So which one should you try first?
If you have a degree, a CV and a LinkedIn profile, and you are in one of the roughly 100 countries on Outlier’s list, try Outlier first. The screening is spread out, the country question is answered before you invest anything, and you are told exactly whose company you are joining. You will not know what it pays until you are inside.
If you do not have a degree, or you want to see real published rates before you commit, try DataAnnotation first, on a morning when you are sharp, and treat the Starter Assessment as the one shot it is. Go in knowing that the rate card disagrees with itself and that the top of each band is a ceiling rather than an expectation.
Doing both is reasonable. Doing both in the same evening is not, for the reason above. And neither is a substitute for looking at what else is open: Outlier alternatives covers twelve platforms in this category, and our full DataAnnotation review goes deeper on that one.
Screened, human-approved, filled ones deleted daily. No account, no paywall.
Frequently asked questions
Is Outlier or DataAnnotation better?
Neither is better across the board. Outlier publishes its country allowlist and names Scale AI as the operator, but publishes no pay figures at all. DataAnnotation publishes a public rate card and asks almost nothing to sign up, but hides its operating company inside a click-through contract and gives you only one attempt at its assessment.
Who owns Outlier?
Its terms of use, effective May 11, 2026, name Scale AI, Inc., a Delaware corporation, and its affiliates including Smart Ecosystem Inc. doing business as Outlier AI. The same terms govern Outlier, Remotasks and the Human Frontier Collective together. Scale AI appointed Francis deSouza as chief executive on July 30, 2026.
Who owns DataAnnotation?
Surge Labs, Inc., a Delaware corporation, named in its terms of use last updated July 14, 2026. That document is reachable only from a link on the signup form, and the name appears nowhere else on the site, including the privacy notice.
Which one pays more?
There is no honest answer, because Outlier publishes no rates at all. DataAnnotation publishes figures from $25 an hour for generalist work up to $150 or more an hour for coding, though those figures differ across four of its own pages. Everything here is as listed by the vendor and is not a forecast of what anyone earns.
Can I retake the DataAnnotation assessment?
No. The vendor states twice that you can take the Starter Assessment once, with no retakes or second chances. Outlier does not publish a retake policy either way.
Do I need a degree for Outlier?
Its FAQ says yes: at least an associate degree. It also asks for a current CV, a LinkedIn profile showing your education and work history, a valid government ID and a mobile phone from your country of residence. DataAnnotation asks for none of these at signup.
Can I work on both at once?
Nothing in either contract prevents it, and many people do. The practical advice is not to attempt both screenings in the same session, because DataAnnotation’s single attempt deserves a fresh head.
Is there any appeal if I get deactivated?
Neither company publishes one. Neither set of terms contains the word appeal or any equivalent, and both reserve the right to deactivate at any time for any reason without notice. Both also require individual arbitration and waive class actions.
Mufy Pachorawala · Founder, getAIwork
My AI scans thousands of AI-job posts (42,827 screened so far) and I personally approve every listing before it reaches the board. I write these articles by the same rules: pay quoted only as listed, and no income promises. Read our editorial rules.
Related: Outlier AI review · DataAnnotation review · Outlier alternatives · Outlier account deactivated · DataAnnotation no response · Surge AI review · AI job market statistics
