Student Says Handshake AI Shorted Him $11,000 for AI Training Work

A college junior says Handshake AI owes him about $11,000 in back pay for hundreds of hours of AI training work, and nearly two dozen current and former contractors told Business Insider the startup underpaid them by hundreds or thousands of dollars each, with disputes over the past six months totaling $64,000.

Joel Thomas, a student who transferred to the University of Mary Washington in the fall, was recruited in August through the career-services app his university had recommended. Handshake AI invited him to apply for AI training work paying $50 an hour or more. He accepted two projects, one at $50 per hour and another at $65 per task, after abandoning earlier spring projects that paid around $17 per hour or less.

His dashboard showed earnings topping $19,000 at one point, he says. He was paid $8,000. Handshake denied his appeals, telling him there was no record of his work during the week of August 17 to 23 — though a Handshake pay statement showed he completed 428 tasks for Project Novel that week and would receive no pay for them.

What caused the payment disputes?

A Business Insider investigation, based on bank statements, pay stubs, and hundreds of email and Slack messages, found several causes:

• Contractors' pay often did not match expectations because Handshake and its clients tracked time and pay differently.
• Dollar figures on contractor dashboards were estimates based on manually logged timers, while the paying client ran a second, invisible timer.
• Handshake compared self-reported hours against the client's timer before paying, but did not tell contractors what counted as billable time.
• Tasks that failed review went unpaid, and confusing or contradictory instructions — some generated with AI coding tools like Claude and Lovable — increased failure rates.
• On some marquee projects, one manager oversaw up to 1,000 contractors, leaving contractor questions on Slack unanswered.

"Final pay reflects the work our systems verify," a Handshake spokesperson said. "Where logged time doesn't match the work, we review it, and where we find inflated time we don't pay for it."

How big is Handshake's AI business?

Handshake, founded in 2014 by three Michigan Tech students as a college career platform, expanded into AI training in January 2025 and rebranded to add AI to its name. CEO Garrett Lord said in August that the AI unit had grown to more than $1 billion in gross annualized revenue. Former staffers said the majority of revenue now comes from AI-training contracts with five clients: Meta, Google, OpenAI, Nvidia, and Anthropic.

Handshake said more than 188,000 contractors have earned over $500 million on its platform. Business Insider first reported in March that dozens of contractors complained of denied payments, and one contractor won a $6,475 judgment in small-claims court. In many cases, the company sent automated messages saying accounts were closed for violating platform rules.

How has the company responded?

In October, after Business Insider sent detailed questions, Handshake introduced pay transparency changes: a new payments dashboard, more explanations for adjustments, an easier dispute process, and a new live timer for work on third-party platforms.

"Our Fellows are the heart of Handshake AI," the company wrote in an October blog post. "As part of being a Fellow you should always know what you earned, how it was calculated and when it will arrive."

Handshake declined to comment on Thomas's case individually. "We take every dispute seriously," a spokesperson said. A dedicated team independent of the project reviews disagreements and can restore all or part of the amount, the company said.

The scrutiny cuts against Lord's marketing. "Most AI gigs = shifting rates, zero transparency," Lord wrote on LinkedIn last year. "Handshake AI = competitive pay, clear expectations, and real dialogue. That's what experts deserve." He has said complaints about late payments at competitors like Scale AI and Mercor contributed to Handshake entering the business.

Thomas posted on Reddit after a low first payment: "This is such a scum thing to do and I am unsure what I should do. Just want to expose them as much as possible because this is ridiculous." Handshake later deactivated his account, citing irregular activity that did not align with platform policies. He remains open to AI training work elsewhere but has given up on recovering the money from Handshake.

The case shows how the invisible-timer pay model now standard across AI-training platforms can collapse trust between platforms and the expert workforces that frontier labs depend on.