Outlier AI Review (2026): Pay, Empty Queues and the Meta Fallout

By Mufy Pachorawala, founder of getAIwork · Last updated: August 25, 2026

Outlier AI is Scale AI’s legitimate expert data platform: it pays weekly and was still onboarding new contributors in August 2026. The defining risk is volume, not fraud. After Meta’s June 2025 investment in Scale, major customers left, task queues thinned, and the platform’s best-known complaint, the empty queue, became more common. Treat it as variable side income.

Key facts

  • Reported pay in 2026 spans roughly $12 to $45 per hour by domain, with specialist projects listed higher. Every figure is per project, changes without notice, and listed is not earned
  • Outlier’s own site cites a Trustpilot rating of 4.1/5 from 3,000+ reviews and claims 700,000+ contributors and $500M+ paid out (August 2026)
  • A $12.5 million class settlement covering California contributors on Outlier and Remotasks is pending: opt-out by September 3, 2026, final approval hearing October 30, 2026
  • After Meta’s $14.3B investment in Scale (June 2025), Google and OpenAI wound down Scale work, and contributors reported thinner queues and rate compression through late 2025
  • 630 AI platform programmes are live on the getAIwork board as of August 22, 2026, screened and human-approved

What is Outlier and how does it relate to Scale AI?

Outlier is the contributor-facing platform of Scale AI, one of the original giants of AI training data. Contributors qualify into domains (coding, maths, languages, generalist writing, and specialist fields) and then work through queues of tasks: rating model responses, writing ideal answers, evaluating code, and red-teaming. Scale consolidated most of its contractor work under the Outlier brand in 2024, and its older crowd platform Remotasks now redirects rejected sign-ups to Outlier.

Onboarding in August 2026, per Outlier’s own FAQ: create an account, verify identity with a government ID and a mobile number from your country of residence, add a resume, then complete unpaid assessments per project before tasking. Payouts run weekly on Tuesdays via PayPal, AirTM or ACH depending on country.

Is Outlier AI legit?

Yes, with an asterisk the size of the empty queue. On the payment question the record is solid: weekly payouts are consistently reported as accurate and on time, and the platform claims more than half a billion dollars paid to contributors. On the treatment question the record is rockier: sudden project pauses, account deactivations with template explanations, slow support, and a wave of worker lawsuits about pay practices filed in late 2024 and early 2025.

Where those suits landed matters for the verdict. A $12.5 million class settlement in McKinney v. Scale AI, covering California contributors on Outlier and Remotasks between December 2020 and February 2026, received preliminary approval this year; the opt-out deadline is September 3, 2026 and the final approval hearing is October 30, 2026. A settlement is not an admission of wrongdoing, but it is the clearest signal yet that the gig-era pay mechanics are being forced to clean up.

Our assessment: legitimate platform, honest payouts, unreliable volume. Join it as one queue among several, never as a plan built on forty hours a week.

How much does Outlier pay in 2026?

Outlier shows each project’s rate before you start, which is the honest way to read it: rates belong to projects, not to you. Synthesising the platform’s listings and 2026 community reporting:

Domain Reported range (2026) Notes
Generalist writing and rating $12 to $18 per hour Most affected by 2025 rate compression and queue thinning
Mid-skill projects $15 to $25 per hour Steady but competitive since layoffs refilled the pool
Coding, legal, medical $22 to $35 per hour Most consistent volume through 2026 per contributor reports
Specialist and expert projects $30 to $75 per hour listed Listed ceilings; access is assessment-gated and intermittent

Two pay mechanics to understand before judging any number. Outlier distinguishes active tasking time from other time, and complex projects can quietly halve an effective rate through unpaid guideline reading and reviews. And when a project ends, its rate ends with it. The listed figure you qualified at is not a salary; it is the price of one queue while that queue exists.

See which AI queues are live for your profile

Answer 6 questions and we match you against 630 live platform programmes on the getAIwork board, by skill, level and country.

Take the 2-minute match quiz →

Free to take · real listings, screened daily · no income promises, ever

Why is my Outlier queue empty? EQ, explained

EQ (empty queue) is Outlier’s signature complaint: an account in good standing with no tasks to do, sometimes for weeks. It has three ordinary causes, none of them about you. Projects end, and the pool empties for everyone at once. Quality audits pause accounts while work is reviewed. And onboarding assessments can leave you qualified for a project that has not started tasking yet.

Since mid-2025 a fourth cause dominates: there are simply more qualified contributors than tasks in several domains, after layoffs and customer departures shrank the work while the contributor pool grew. The practical response is the same as on every pooled platform: keep two or three platforms qualified, watch for new project intakes, and put your hours wherever tasks currently exist. Our platform comparison explains the pool-versus-project mechanics that drive this.

Outlier work is queue-based, and queue volume varies sharply from week to week
An empty queue is a signal, not a sentence

When your queue dries up, the fix is a second qualified platform. See which programmes are taking people this week.

Find your match in 2 minutes →

Free to take · real listings, screened daily · no income promises, ever

What actually changed after the Meta deal?

June 2025 reshaped the whole platform. The sequence, from reporting at each step:

When What happened Effect on contributors
June 13, 2025 Meta invested $14.3B for 49% of Scale; CEO Alexandr Wang left for Meta Uncertainty, then customer flight
June 14 to 18, 2025 Google (reportedly Scale’s largest customer) and OpenAI moved to wind down Scale work Less client work entering the pipeline
July 2025 Scale cut 200 employees and 500 contractors Laid-off contractors refilled remaining queues
Late 2025 Generalist writing programmes wound down; investigations reported baseline rates compressing Fewer, lower-paid generalist tasks
2026 $12.5M class settlement moves forward; Outlier keeps onboarding in coding, STEM, languages Platform stabilises around skilled domains

The takeaway is not that Outlier is dying; it is that the easy generalist era ended. The domains still hiring and tasking in August 2026 skew toward coding, STEM, languages and credentialed specialisms.

Who does Outlier still make sense for in 2026?

Worth joining if you bring a skill the remaining queues want: code you can review, a strong STEM background, a in-demand language pair, or a licensed profession. Contributors in those domains report the steadiest volume and the best rates on the platform. Worth skipping if you need predictable weekly income, or if generalist writing is the only queue you would qualify for; that specific work is thinner and cheaper than it was in 2024, and open-intake alternatives may serve you better while you build a specialism. Either way, qualify on more than one platform before you need to.

Outlier is one queue. The board shows them all.

1,234 live, human-screened AI work listings, updated daily, with 139 marked beginner-friendly.

See the live board →

Free to take · real listings, screened daily · no income promises, ever

Frequently asked questions

Is Outlier AI a scam?

No. It is Scale AI’s contributor platform, pays weekly via PayPal, AirTM or ACH, and claims over $500M paid out. Complaints concentrate on empty queues, sudden pauses and slow support rather than missing payments, and a $12.5M class settlement over past pay practices is pending.

How does Outlier pay you?

Weekly, on Tuesdays, via PayPal, AirTM or ACH depending on your country. Each project shows its rate before you start. Rates are set per project, differ by domain and country, and end when the project ends.

Why did my Outlier queue suddenly go empty?

Usually a project ended, an audit paused your account, or a qualified project has not started tasking. Since 2025 there is also simple oversupply in some domains: more qualified contributors than tasks. It is rarely personal, and it is the main reason to keep a second platform qualified.

What is the Outlier lawsuit about?

A wave of worker suits from late 2024 alleged misclassification and pay violations. McKinney v. Scale AI produced a $12.5M class settlement covering California contributors on Outlier and Remotasks from December 2020 to February 2026, with final approval scheduled for October 30, 2026. A settlement is not a finding of wrongdoing.

Is Outlier still hiring new people in 2026?

Yes. As of August 2026 its site actively recruits for coding, STEM, languages and generalist domains, and new sign-ups complete ID verification plus unpaid project assessments. Acceptance does not guarantee task volume, which is the distinction that matters.

Do I need a degree to work on Outlier?

Depends on the project. Requirements range from no degree for some generalist queues to advanced degrees for expert projects. Skills that unlock the steadiest 2026 queues are coding, STEM, and in-demand language pairs.

Which is better, Outlier or DataAnnotation?

Different failure modes. DataAnnotation has harder entry (one-shot assessment, 2.6% claimed acceptance) but strong listed generalist rates; Outlier is easier to enter but queue volume swings harder. Workers who sustain this category typically qualify on both and follow the tasks.

Can I rely on Outlier as a main income?

The volume risk makes that unwise. Weekly income can go from full to zero when a project ends, through no fault of yours. Treat it as variable side income, keep parallel platforms qualified, and never plan fixed expenses against queue-based work.

Mufy Pachorawala

Mufy Pachorawala · Founder, getAIwork

My AI scans thousands of AI-job posts (38,434 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: DataAnnotation.tech review · DataAnnotation response times · Data annotation jobs compared · Get paid to train AI

More from the blog