Companies will pay you to catch chatbots making stuff up. Here’s how much, where to apply and the fake ‘AI trainer’ jobs to dodge.

September 7, 2026

By Kim Komando

I get this question a lot on the show. “How can I make money at home with AI?”

Chatbots flub facts every single day, and somebody gets paid to catch every one of them. Might as well be you.

💰 To err is machine, to profit is human

Here’s the gig: AI companies need humans to grade the robots’ homework. You read a chatbot’s answers, flag the wrong ones, fix clunky writing or compare two responses and pick the better one. No coding needed for most of it. Just a sharp eye and solid English.

And the pay is real. General projects run $15 to $40 an hour. Nurses and folks with medical training pull about $85 an hour reviewing health answers. Doctors, lawyers and PhDs can clear $100-plus.

Here’s where to apply. Type these addresses yourself:

• DataAnnotation.tech: Most folks start at $20 to $25 an hour, coders $40 and up. Pays weekly through PayPal.
• Outlier.ai: Writing and fact-checking projects, with specialist work up to $75 an hour.
• Alignerr.com: Credentialed pros (health, legal) can hit $95 an hour.
• Mindrift.ai: Advertises $15 to $100-plus, depending on your expertise.

FYI, the fine print: You’re a freelancer, work comes in waves and there’s an unpaid skills test to get in. It’s side money, not a salary. Don’t quit your day job. Do check your email in line at the grocery store.

🚨 Fakers gunning for your wallet

Scammers watched this gig get hot and built a trap called the task scam. It starts with a text or WhatsApp message offering easy “app optimization” work. The first little tasks pay out a few bucks, right on time. Then comes the twist: Deposit your own money, usually crypto, to “unlock” the higher-paying level. That deposit is gone forever.

Three rules keep you safe:

1. Real platforms never text first. You apply to them, never the reverse.
2. Never pay to get paid. No deposit, no “training fee,” no crypto. Ever.
3. Skip the ads. Type the site address into your browser yourself. Sponsored look-alikes are how crooks catch you.

You’ve spent your whole life correcting know-it-alls for free. Fair’s fair. Time the know-it-all paid up. Finally, being a nitpicker isn’t a character flaw. It’s a billing category.

🎯 Send this to someone who corrects everyone’s grammar for free.

📱 Text-worthy: Companies pay people to catch chatbot mistakes: $15-$40/hr to start, ~$85/hr with a medical background, $100+ for lawyers and PhDs. Legit sites: DataAnnotation.tech, Outlier.ai, Alignerr.com, Mindrift.ai. Apply directly. Never pay a fee. Any “job” that texts you first is a scam.