BreakingEven

Are the AI Platforms Going Full Hunger Games?

TL;DR: The biggest platforms have started staffing single projects with tens of thousands of contractors at once — Mercor put 20,000+ in one Slack channel for Project A; Handshake's new project is climbing fast. That's not generosity and it's not a mistake. It's a model: hire a huge crowd, pay only for the minutes anyone actually works, drop a giant pile of tasks with no warning, and let the crowd swarm it until it's gone. It clears the backlog in minutes. For you it means high pay you can rarely log, no real voice, and a "thank you for your service" that one random day turns into an email that says paused — more to come. The money is real. So is the exit. This is about deciding, going in, whether those are rules you can live with.

Are the platforms moving to a Hunger Games business model for the people they contract?

I've watched this trend take shape for a few months now, and I don't think it's an accident anymore. Platforms are standing up projects that involve tens of thousands of contracted workers at the same time. Mercor brought 20,000+ contractors into a single Slack channel for what people only refer to as Project A. Handshake has its own new project doing the same thing on the same trajectory.

Their newest project — workers are calling it Hydra — started around May 29 and was already past 4,000 contractors within days. Last I heard, it was north of 10,500, and still climbing, which tracks. Mercor went from 2,000 to 22,000+ in its first month.

So here's the real question, the one nobody at the top is asking out loud: why would anyone want that many people clamoring for the same work at the same time?

My bet on what's really happening

I'm a betting man. So let me tell you what I'd put money on.

I'd bet Double Top Dollar — and I love that game — that the client called up and said some version of: "Hey, we have to get Claude (or Copilot, or Gemini, or GPT) to learn this thing, and we need it to learn it NOW. Money is no object."

And that's it. BOOM.

Mercor and Handshake don't know the schedule. They don't know when the data's coming from the client. They just know it has to get done now. So somebody brainstormed the obvious problem: how do we process 100% of the work, so fast, that there's never a backlog?

And the answer they landed on is the one we're all living inside right now. Hire 20,000 people. Pay an eye-watering hourly rate. Those people will circle the work like it's the last helicopter out, and the second a big dump of tasks lands, the crowd is already there to devour it.

I'm sure someone in that room said that's too many, it'll be a nightmare to manage, it'll cost a fortune. And someone else said: nonsense. They only get paid for the minutes they actually work. Stick them all in one Slack channel with an AI for answers — no extra staffing on our end. And if a backlog ever builds, all we do is send a text that says there's work, and I'll bet you cash money enough of those 20,000 show up that the backlog is gone in twenty minutes.

Let the games begin. And may the odds be ever in your favour.

Except they aren't, are they?

Here's where the framing stops being cute.

The two coordinators running a 20,000-person project are not going to hear your individual voice. There's almost no chance they'll even see your Slack post — because the instant you hit send, thirty-three other people have already posted behind you. The channel isn't a conversation. It's a stampede with a scrollbar.

You can watch it happen in real time: people refreshing for ten minutes straight and landing nothing, task drops that are gone before the page finishes loading, a channel scrolling faster than anyone can actually read it. Nobody in there is exaggerating the speed — that is the speed.

We discussed this in May when Mercor's project started taking off. Hydra isn't a new idea. It's confirmation of a business model that's being tested by some of the biggest platforms that are out there.

The rules of the game

So you know the rules now. The job is to decide whether they're rules you can live with, or whether they're enough that you need to take a step back.

Lay them out plainly, because the platforms won't:

  • You get paid only for the minutes you can actually grab work. The hourly rate is real — and for Mercor's top tiers it's genuinely a Benjamin or more an hour — but the rate is only worth what you can log. A $100/hour project you can touch for 45 minutes is a 45-minute job.
  • Availability beats everything. This is generalist crowd work. Being online and ready when the dump lands matters more than the headline number. The people who do well aren't the smartest in the channel; they're the fastest to the button and the most relentlessly there.
  • You have no voice. Not because anyone's cruel — because math. Two people cannot hear twenty thousand.
  • It ends with no warning. One random day — maybe not tomorrow, maybe not next week, but one day — you get an email that says the project is paused. More information to come. No explanation. No cause. No thank you.
  • The rules can and have changed — because the client says so. A number of different tasks have been introduced and time limits have been placed on work. If you don't meet those times, then your work disappears and you've just logged 45 minutes of unproductive time.

That last one is the whole thing. So let me tell it the way it actually feels.

The man with the groceries

Imagine someone tells you: one day you're going to help a person cross the street, and when you do, it's going to hurt more than you can believe. But if you refuse to help, you don't get to make any money here at all.

So you help. And it's fine. You help someone across, and then another, and another. After enough of them you've forgotten there was ever supposed to be pain. You haven't thought about it in years.

Then one ordinary afternoon you're helping a random man with his groceries, walking him across the street, and the second you reach the other side he whacks you in the kneecaps and walks off. No thank you. No goodbye. Just gone.

That's the deal. That's effective-immediately, without cause, without a goodbye. The platforms have built a model where that ending isn't a betrayal — it's a feature. The whole point of hiring 20,000 people is that any one of them is disposable the instant the platform decides, and for a reason you will probably never know. If the odds have been ever in your favour, the end will be just as abrupt for you, and it will end with a random email saying the project has been paused.

So can you live with it?

This is the part I can't decide for you, and won't pretend to.

Here's the thing, though. If you can compartmentalize — if you can separate the personal feelings, even the moral standing, and make this purely about being handed a temporary opportunity to make real money, knowing full well that one random day it ends with a whack to the kneecaps — then this can absolutely be worth playing. The money is not fake. People clear serious weeks on these projects.

But you have to play it like what it is. A few rules I'd hand anyone walking into a mass-hire project right now:

  1. Work every drop like it ends in two hours. Not tomorrow. Not after lunch. The minute you start expecting the $4,000 week to roll in on schedule is the minute you're exposed when the train stops.
  2. Never spend it before it clears. This is non-negotiable across every platform I track, and doubly so here. The pause email and the payment delay tend to arrive in the same season.
  3. Treat the rate as a ceiling, not a wage. Plan your life around what you can realistically log, not the billboard number.
  4. Read the documentation like your check depends on it — because it does. Every one of these projects buries the rules in onboarding. The people screaming "I didn't know that was a rule!" were usually paid to read the rule.
  5. Keep your own receipts. Hours, screenshots, payment confirmations. When there's no voice and no goodbye, your records are the only version of events you control.

None of that changes what the model is. It just lets you walk in with your eyes open instead of finding out on kneecap day.

The platforms have decided the most efficient way to never have a backlog is to keep a standing army of contractors hungry and circling. That's the game they are testing out now. You don't have to be mad about it to refuse to be surprised by it.

May the odds be ever in your favour. Just don't forget who owns the arena.

Read Next

Is $100/Hour at Mercor Actually Possible?
Mercor advertises rates up to $200/hour and hired thousands of generalists for one of its biggest projects. So is $100/hour real? Based on what workers are saying inside the community: yes — but it's a unicorn, and only if you're online when the work drops. Here's the honest picture, including the payment caution everyone learns the hard way.
Handshake AI Review: The $80/hr Platform That Keeps Offboarding People
Handshake AI's Project HH workers are reporting 20–50% of earned pay. Wage theft allegations, mass arbitration organizing, and platform investigation rumors as of May 2026. Updated with the current situation.
DataAnnotation & Outlier AI Work Drought: Why Tasks Disappear and How to Survive It
Is your DataAnnotation queue empty? Discover the lifecycle of the AI gig worker, from the first payout to the inevitable 'No Tasks Available' screen, and learn how to survive the bust.
Outlier AI Empty Queue: Fired, Glitched, or Just Paused? How To Tell
Is it a glitch, or were you fired? We break down the 'Silent Severance' of AI training platforms like Outlier and DataAnnotation, and why you have to stop blaming yourself when the work suddenly stops.

What workers are talking about

Loading the latest threads…
Was this helpful?

Comments

Leave a comment

Your email won't be published.

Joshua Drake has worked on AI training platforms for over four years, tracking earnings, sentiment data, and platform stability across Outlier, DataAnnotation, Alignerr, and others. He has a degree in data analytics and runs this site, breakingeven.online and the sentiment analysis used to derive a sense of what is happening in a world often hiding in the shadows.