The Job Board Is Lying to You | NaN's Blog

🇺🇸Switch languageToggle theme

2026-07-30 AI

The Job Board Is Lying to You

0 views

TLDR:<br>27% of LinkedIn job listings don't exist. AI companies like micro1, DataAnnotation, and Mercor flood job boards with vague postings as data harvesting funnels. Even traditional-looking jobs are annotation work in disguise.

Photo by julien Tromeur on Unsplash

The Data Funnel

I spent two years working for myself. When I finally opened LinkedIn to test the waters, I searched for my skills and filtered by the past week.

The results were garbage. Hundreds of links pointed to companies I had never heard of. Mercor. micro1. Outlier. DataAnnotation.

They all wanted an "AI Trainer" or an "RLHF Evaluator." No salary bands. No team details. Just vague copy about the future of artificial intelligence. I kept scrolling. The main feed was worse. Finding an actual open position felt like digging through a landfill.

Did these companies ever intend to hire anyone? I doubt it.

The Play

The business model is beautifully cynical.

They post a broad job listing. They make it sound revolutionary. Applicants swarm the posting. Within hours, candidates receive a "skills assessment" that takes three hours to complete. You finish the test. The company promises to be in touch. You never hear from them again.

*:last-child]:mb-0">This is a data acquisition pipeline. Your test work and interview recordings become free training data. Mercor's TOS writes it right into the terms - they own commercial rights to applicant contributions. micro1 tracks your audio, video, and browser tabs during interviews.

These are data brokers wearing the skin of recruitment firms. Mercor raised $100 million from Peter Thiel. They ran 100,000 interviews and spiked their revenue to $500 million in 17 months . Nobody knows how many people they actually hired. Meanwhile, AI job postings are up 134% while real tech hiring is down 34% .

The Disguise

You learn to ignore the obvious spam. The real danger is the disguise.

I applied to a posting that looked completely normal. Traditional title. Real company. Valid salary range. I applied. I got an interview.

Nobody showed up. An AI interviewed me. I recorded responses into a camera, answering prompts alone. Then came a one-hour test. An AI scored it. The result email was sent by an AI.

No human was involved at any point. Not in screening. Not in interviewing. Not in grading. Not in communicating the result.

*:last-child]:mb-0">The posting looked like a real job. The interview was a recorded video session. The test was annotation work. The entire process existed to extract an hour of my time - and whatever data my face, voice, and answers generated - without a single human on the other end.

This is the pattern. A listing that looks legitimate. An interview with no interviewer. A test scored by software. A rejection email written by a language model. You never talk to a person. You never get feedback. You just give them data.

If a coding job asks you to grade chatbot text, walk away. If you find yourself talking into a camera for a one-way interview, you are generating training data. If the entire hiring process has zero humans in it, the process is not hiring. It is harvesting.

Ghost Jobs and the Doom Loop

Then there are the ghost jobs. Openings that simply do not exist.

Zero funding. Zero headcount. Companies post them to harvest resumes or project fake growth. A Columbia Law Review study found 48% of tech job listings are complete phantoms. Clarify Capital found 1 in 3 employers admit posting jobs with no intent to fill them. 68% of companies believe ghost jobs have positive revenue impact. Employers do this intentionally because it suppresses wages and builds pipelines.

LinkedIn's overall ghost rate is 27%. More than a quarter of every listing on the platform does not exist.

The hiring market is trapped in a doom loop. Greenhouse CEO Daniel Chait describes the cycle: AI tools let applicants spam hundreds of jobs a day - 254 applicants per posting, up 412% . Companies panic and deploy AI screening to auto-reject people. 75% of companies let AI reject candidates without human oversight. Candidates fight back by hiding prompt injections in their resumes. 91% of recruiters caught candidate fraud in 2025. Trust is completely dead - 70% of hiring managers trust AI, only 8% of job seekers call it fair. The humans on both sides lost control of the machine.

The Reality

Do not wait for the government to save you. Ontario passed a law - Bill 190, effective January 2026, the strongest in North America. New York's S8877 passed the state Senate in April 2026. The Texas Attorney General is investigating LinkedIn for profiting off fake listings while selling Premium subscriptions. LinkedIn responded by adding a verified badge. Half the jobs on the platform still lack it.

No US state has enacted enforceable law as of July 2026.

I stopped trusting the board. I read the...