How to spot legitimate remote AI work

There is real, well-compensated remote work at frontier AI labs right now. There is also a growing ecosystem of scams that mimic it. Advance-fee frauds, fake recruiter pitches, cryptocurrency-adjacent grifts, and outright identity theft schemes all use the same surface vocabulary as legitimate expert work. For a professional in Pakistan, Nigeria, India, Egypt, or the Philippines who has heard the phrase "AI training data" for the first time this year, being able to tell the two apart is table stakes.

This piece is a practical guide. It is written in the register a professional would want, not the tone of a defensive FAQ.

What legitimate expert AI work looks like

Legitimate work has consistent shape across every frontier lab that does it. Whether the specific project is medical annotation, legal evaluation, or mathematical adversarial testing, the surface signals are the same.

You are contacted through a real channel. Not a WhatsApp message from a stranger. Not a Telegram invite. Not an email from a Gmail address impersonating a company. Legitimate outreach happens through your professional profile on a platform you signed up to, or through a company address on a domain you can verify by hand.

The work is described in plain terms. You are told what field the project covers, roughly how many hours are expected, what deliverable you will produce, and what platform you will use to produce it. You are not told the work is "confidential" and cannot be described.

You are paid, not asked to pay. You never send money, never buy equipment, never pay a "verification fee," never transfer cryptocurrency to prove you are real. Every real project pays you in USD on a schedule that is written down in advance.

There is a written agreement. A short work-for-hire document, or a click-through NDA and terms sheet, is standard. It names the specific work, the rate, the deliverable, and the payment terms. It is not a twenty-page mystery document you cannot read.

The compensation is real but not ludicrous. Rates for expert work are on Western scales, in USD. For a physician in Karachi that is a large multiple of local hourly compensation. It is not a hundred times that. If the promised rate is 500 USD an hour for entry-level annotation, that is a scam.

What scams look like

Direct-message outreach that skips your public credentials. A stranger contacts you on WhatsApp, Telegram, or Facebook Messenger, saying they represent a large AI company, and can offer you work immediately if you send a small "setup fee," "wallet address," or "identity verification document."

Vague deliverables plus urgency. You are told the work is "training AI" or "labeling data" without specifying what field or what kind of data, and you are told the opportunity closes soon.

Payment before work. You are asked to buy specific equipment, register on a specific platform that requires a deposit, or wire money to demonstrate seriousness.

Cryptocurrency involvement. Legitimate frontier AI work is paid in USD via bank transfer, Wise, Payoneer, or direct payroll systems. If a project pays only in USDT, only in a specific memecoin, or only through a smart contract you do not fully understand, walk away.

Impersonation of a real company. Someone contacts you claiming to represent OpenAI, Anthropic, Google DeepMind, Scale, Micro1, or another named lab. Real recruiters at those companies do not send unsolicited WhatsApp messages. Verify the sender by looking up the person on the company's official channel. If you cannot verify, it is a scam.

Any request for your government ID before you have a written engagement. ID verification is standard once you are onboarded to a real project, but never in advance of any written agreement.

How Remotebridge fits

Remotebridge is a discovery service, not a staffing agency. We do not employ you. We do not take money from you. We do not send you unsolicited direct messages. Every role in our catalog is a live posting at a frontier AI lab that has published it publicly, with a referral link that lets us track successful placements without touching your compensation.

If you sign up, upload a resume, browse the catalog, and click Apply on a role you like, the referral link takes you directly to the lab's own application flow. From that point on, everything happens on the lab's platform: their contract, their onboarding, their payments to you. We are not in the middle of that.

This structure is deliberate. It means we have no way to defraud you and no incentive to. Our payout, when it comes, comes from the lab, not from you.

Practical rules

If you take nothing else from this piece, take these five.

1. Never send money to receive work. Never.
2. If a stranger contacts you on WhatsApp, Telegram, or Facebook offering AI work, treat it as a scam by default.
3. Verify any recruiter by looking them up on the company's official channel before responding.
4. Read every document you sign, however short. If you cannot read it, do not sign it.
5. Payment should be in USD through a real payment rail (bank, Wise, Payoneer). If it is not, walk away.

Real remote AI work exists. It pays well. It is not sent to you unsolicited in a Telegram invite. If in doubt, browse the open roles directly and apply through channels you can verify.