On October 5, 2026, a counter on micro1’s own reviews page read 2,085,933 candidate reviews. The same morning, its public job board showed 266 open roles.
Those two numbers explain most of what people say about micro1 online. Getting an interview with Zara, its AI interviewer, takes a few clicks. Getting paid work is a separate step, and micro1’s own FAQ says certification doesn’t guarantee immediate placement on a project.
micro1 is legit. It’s a funded San Francisco company with real clients, real contracts and a published pay schedule. Our verdict is still Caution, and the reasons aren’t about fraud. The recording of your interview becomes micro1’s property. It can be used to train micro1’s own AI models unless you opt out. And the contract lets the company end your work with no notice.
Last checked: October 5, 2026. How we checked this is explained at the end.
Most people only ever meet the interview
micro1 began as an AI recruiting business and now sells human expertise to AI labs. Experts write, grade and correct model answers. Generalists record video for robotics training. Forbes reported on September 22, 2026 that founder Ali Ansari, 25, had raised more than $100 million at a $4 billion valuation. A month earlier, TechCrunch put its gross annual run rate at $500 million.
So the company is real, and so is the money. How much of it reaches the average applicant is a different matter.
Take that counter. micro1 labels it a live candidate experience tracker, with an average of 4.33 out of 5. It doesn’t say exactly what gets counted, and the total may include repeat attempts and interviews Zara runs for client companies. Even so, two million ratings set beside a few hundred openings tells you where most people’s micro1 story begins and ends.
Trustpilot shows the same pattern. micro1 scores 4.7 there from 10,655 reviews, 83% of them five-star. We read the 20 newest on October 5, all posted in the previous 24 hours. Sixteen described the AI interview or assessment. Only two described paid work: one praised steady work and good pay, and the other said the reviewer had been wrongly removed from a project. Trustpilot also notes that micro1 invites reviews and hasn’t replied to its negative ones.
That score mostly rates a conversation with a bot. It tells you little about pay, how long projects last, or what happens when the work dries up.
Your Zara interview belongs to micro1
The fear is spelled out in a one-star Glassdoor review from April 2025: that the interview exists to collect your face, voice and expertise for free. (Glassdoor’s overall score for micro1 was 4.4 when we looked.) micro1’s documents don’t support the harshest version of that claim. They do confirm more of it than you might expect.
We read the Candidate Terms and the Candidate Privacy Policy, both last updated July 21, 2026. Here’s what they say.
| Question | What the documents say |
|---|---|
| Who owns the recording? | micro1. Section 8 of the terms gives it all rights in the interview, its content and your results, and you assign yours to it. |
| Can it train AI on it? | Yes, micro1’s own models, using de-identified audio, video, transcripts and scores. The policy says this isn’t shared with AI model companies to train theirs. |
| Is it sold? | The policy says candidate information isn’t sold or used for targeted advertising. |
| Who else sees it? | Clients can be shown your candidate profile. Vendors that process video and run language models handle data on micro1’s behalf. |
| How long is it kept? | No fixed period. It’s kept for as long as micro1 considers necessary. |
| What if something goes wrong? | Liability is capped at $100. Disputes go to confidential JAMS arbitration under Delaware law. |
| What if you break interview rules? | Outside help and screenshots are banned. micro1 can bar you and report the violation to prospective employers. |
Two loose ends. The privacy policy says de-identified data can be used and disclosed for any purpose, then says in the same section that it isn’t passed to AI model companies for training. And micro1’s help library says ID checks are handled entirely by an outside provider and that micro1 never sees or stores your documents, while the privacy policy lists government ID information and photos among the data it may collect. Ask [email protected] which one applies before you scan a passport.
How to switch off AI training on your interview
None of the reviews we compared this one against mention it, but micro1 runs a self-serve privacy page, linked from the policy. It opened without a login when we tried it on October 5, 2026.
- Go to talent.micro1.ai/delete-account. Despite the address, deleting isn’t the only choice.
- Under Manage your privacy & data, pick Opt out of model training. The page describes it as: “Opt out of future training of micro1 models on your interview data.”
- Click Next and follow the prompts. We stopped at this screen and didn’t submit a request, so we can’t describe the steps after it.
- The same page offers Access your personal data and Delete your account. For anything else, the policy gives [email protected].
micro1’s privacy and data page, with the model training opt-out selected. Screenshot taken October 5, 2026.
Read the word “future” carefully. The opt-out doesn’t promise to undo training that has already happened, so use it as early as you can.
270 hourly listings, counted
The board’s header said 266 roles. The feed behind it returned 304 listings on October 5, 2026. We set aside 31 salaried staff jobs and 3 with no hourly rate, which left 270 hourly contract roles.
The median listing runs from $50 to $100 an hour. The big numbers are real: 142 listings (53%) top out at $100 or more. But only 65 (24%) start there. At the other end, 121 (45%) start below $50 and 53 (20%) start below $30. In half of all listings the top of the range is at least double the bottom, so the figure in the ad says little about your figure.
| micro1’s role tag | Listings | Median range per hour | Start under $30 |
|---|---|---|---|
| Expert | 118 | $90 to $135 | 0 |
| Specialist | 53 | $60 to $100 | 8 |
| Generalist | 87 | $30 to $60 | 38 |
| Robotics | 5 | $10 to $15 | 5 |
Seven untagged listings aren’t shown. The lowest rates sit in video work for robotics data. An at-home video recorder role for Latin America listed a flat $6 an hour, the US version $10 to $15, and two video annotation roles $8 to $10. Forbes reported micro1 paying $50 to $90 an hour to annotate robot videos, so the rate clearly depends on the role and on where you live. At the top were attorney roles at $140 to $400.
The contract in six lines
Pass the interview, get matched, and you’ll be sent an expert agreement. micro1 summarises it in its public library, and the summary is blunt.
- You’re an independent contractor. Taxes, insurance and equipment are yours.
- Everything you produce is work for hire and belongs to micro1.
- You confirm you aren’t currently employed by a foundational AI model company.
- You can quit with at least 5 business days’ written notice. micro1 can end the agreement immediately.
- Disputes go to binding JAMS arbitration. You waive jury trials and class actions.
- For a year afterwards you can’t solicit micro1’s staff or other contractors.
Pay runs twice a month through Deel. The payroll policy says work from the 1st to the 15th is paid around the 20th, and work from the 16th to month end around the 5th, with US contractors sometimes a business day later. Hourly projects are timed with Hubstaff’s desktop app, and only time spent actively working on the project counts.
We found no pattern of unpaid work in the reviews we read. The sharper risk is the one the contract spells out. A client cancels a project, or you’re removed from one, and the income stops that day.
Messages that claim to be from micro1
A company this visible gets impersonated. Every genuine page we used sat on micro1.ai or one of its subdomains: talent.micro1.ai for accounts, jobs.micro1.ai for applications, payroll.micro1.ai for pay and library.micro1.ai for help. micro1’s own addresses ended in @micro1.ai, and contracts arrive through Deel.
Nothing in the terms or the library pages we read mentions a fee to apply, interview or get certified. Treat any request for money, gift cards or crypto as fake. The same goes for a “recruiter” who moves you to Telegram or WhatsApp and asks for bank logins. Our Handshake AI review covers how the same fake-recruiter trick works on another platform.
Apply, but on these terms
micro1 suits people with a real specialism (law, medicine, finance, engineering, a less common language) who want flexible contract income and already have another source of it. It’s a poor fit if you need predictable hours. It’s a worse one if recording a video interview that someone else then owns bothers you.
- Read the two candidate documents before the interview. They’re short.
- Decide whether to opt out of model training, and do it early.
- Find a listed role that matches you first. A certificate on its own isn’t work.
- Budget on the bottom of the pay range, not the top.
- Keep your own log of hours and of every approved pay period.
If you’re comparing platforms, our reviews of Mercor, Alignerr, DataAnnotation and Outlier ask the same questions of each.
How we checked this
On October 5, 2026 we read micro1’s Candidate Terms, Candidate Privacy Policy, experts FAQ and public library pages on the expert agreement, payroll, Hubstaff and ID checks. We opened its reviews page, its privacy and data page and its job board, and pulled all 304 listings from the board’s public feed to work out the pay figures. We read the 20 newest Trustpilot reviews and one Glassdoor review, and the Forbes and TechCrunch reports linked above.
We didn’t apply, sit a Zara interview, sign an agreement or get paid, and we didn’t submit an opt-out request. The pay figures are advertised ranges, not earnings. We can’t audit micro1’s review counter, and the contract points come from micro1’s own summary of its agreement, not the full document.
Prices, plans and features change often. Check the official site before you pay. Datalevo isn’t affiliated with or paid by any company mentioned here. Research and drafting were AI-assisted, and every figure was checked against the linked sources. Our full process is on the How We Review page.
FAQs
Is micro1 legit or a scam?
micro1 is a legitimate, venture-funded company in San Francisco that hires contractors to train and test AI models. The cautions are about its terms: it owns your recorded interview, and it can end your contract without notice.
Does micro1 use my interview to train AI?
Its Candidate Privacy Policy reserves the right to train micro1’s own models on de-identified interview audio, video, transcripts and scores. You can opt out of future training on the privacy and data page at talent.micro1.ai/delete-account.
How much does micro1 pay per hour?
Of 270 hourly listings on its board on October 5, 2026, the median range was $50 to $100. About a quarter started at $100 or more, and a fifth started below $30. These are advertised ranges, not guaranteed earnings.
When does micro1 pay?
Twice a month through Deel. micro1’s payroll policy says work from the 1st to the 15th is paid around the 20th, and work from the 16th to month end around the 5th of the next month.
Does passing the Zara interview guarantee work?
No. micro1’s FAQ says certification makes you eligible for projects but doesn’t guarantee immediate placement. You still need to be matched to a project or apply to an open role.
