AI Hiring Black Hole: Why the Job Market Is Broken

Article and Analysis by Iaroslav Belkin, founder of Belkin Marketing

Last week, expecting to help a few people, I posted on LinkedIn offering direct introductions for displaced executives from BitMEX and BitMart. What arrived was a flood, and not even from junior employees working their way up! Senior business development professionals, operations leads, regional heads who had spent years building market presence all over the globe, suddenly stranded in a hiring market that has no mechanism to evaluate them.

Two Closures, One Week, AI Hiring

BitMEX announced it would permanently shut down on September 23, 2026, after eleven years. Four days later, BitMart announced an orderly wind-down effective January 31, 2027. Two oldest crypto exchanges went in a single month, each citing some version of market conditions and regulatory pressure.

Sure, AI did not close these exchanges. Collapsing retail volumes, rising compliance costs, and years of over-hiring into a market that normalised unsustainable growth did. But the people now on the market from these firms are walking into a hiring infrastructure that was already broken before their employers folded, and has since been made substantially worse by a different AI story entirely.

More than 165,000 roles have been cut globally in the first seven months of 2026 under AI-related rationales, across technology, finance, logistics, and media. Amazon, Accenture, and dozens of others have cited efficiency gains from AI as the justification for headcount reduction. AI washing is the corporate excuse of the decade: cut the humans to fund the GPU clusters, and call it transformation.

The Bot-vs-Bot War in Job Posting

Here is what happens on the other side of the hiring funnel, happening now at scale.

AI auto-apply tools allow a single candidate to submit between 80 and 1,000 tailored applications per day. LazyApply’s Ultimate tier caps at 1,500 daily submissions. FastApply’s Elite plan processes 1,000 applications a day at $49 per month. The average U.S. job seeker sends 84 applications before landing an interview. These tools claim to do that in two hours. And some of them actually can.

The HR response: automated classifiers. Workday, Greenhouse, and Lever all shipped AI-content detection systems in late 2025, specifically designed to downgrade resumes matching GPT default patterns. 49% of hiring managers automatically dismiss applications they identify as AI-generated, according to a Resume.io survey of 3,000 hiring managers. 62% reject AI-generated content specifically when it lacks personalisation. 82% of those same companies use AI to screen the applications they receive.

Read that last sentence again.

The trust collapse runs deeper than classifiers. A viral image of a woman supposedly getting a company tattoo for a job interview spread across LinkedIn as proof of a dystopian hiring market — until the woman in the photo came forward to explain her image had been used without permission and she had never even applied for any job! The story warning about the collapse of trust became an example of it!

The companies using AI to filter applications are penalising candidates for using AI to write them! An irony after irony that writes itself.

As a result of both sides defending themselves against the other, we now have a system running on spam. For every 180 people who apply to a typical role, approximately five get an interview, down from roughly one in eight a decade ago. Software engineering roles at major companies regularly pass 2,000 applications before screening begins. Only 3% of applications result in an interview, according to CareerPlug data covering more than 10 million applications. When both sides of the funnel are automated, trust disappears.

What the MIT Research Actually Shows

There is a data point that gets quoted constantly regarding this and almost always misread.

An MIT and NBER randomised controlled trial across 480,948 job seekers found that using AI to edit human-written prose increases hire rates by 7.8% and wages by 8.4%. That number travels fast. What travels less far is the other half: using AI to ghostwrite applications from scratch triggers the 49% auto-dismissal. The difference between AI-assisted and AI-generated is not detectable to the casual reader. It is very detectable to a trained classifier, and to a recruiter who has read 2,000 applications that all open with “results-driven professionals with a proven track record.

The AI resume paradox, as it has come to be called, means the tool that helps you if used correctly gets you rejected if used the way most people use it. 29% of job seekers are currently submitting AI-generated keyword-stuffed applications under the mistaken belief that this beats the ATS. It does not. It floods the system further and buries everyone else.

The Human Cost Nobody Is Measuring

The executives messaging me after the BitMEX and BitMart announcements had something in common beyond the industry. Most of them had given an unreasonable amount to their roles. Not metaphorically. Physically. Years of time zone arbitrage, 14-hour days across Hong Kong and European and American hours simultaneously, health deferred in exchange for equity and market position that no longer exists.

“My body gave out before I did.” I heard versions of that sentence more than once. It is not a metaphor. An experienced customer success professional documented applying to 10,000+ jobs over two years, draining savings, taking on credit card debt, and ending up in hospital with chest pain the doctors eventually attributed to stress.

These are not people who need to be told to work harder. They need a hiring infrastructure capable of evaluating what they actually built, and right now that infrastructure does not exist. It is running a spam filter.

Surviving this environment requires moving away from public job boards entirely, building direct peer networks, and creating an owned professional presence that algorithmic filters cannot touch. The full analysis on what that looks like in practice is in our guide on AI job loss protection. The short version: the people who will get through this without losing months to the black hole are the ones who stopped competing in the bot-vs-bot arena before it consumed them.

Does that mean, it’s who you know not what you know? Irish Tech News took that question to Ireland-based recruitment expert Owen Healy and that’s what he had to say: ‘Absolutely not. But, it’s an employer’s market so having somebody on the inside advocating for you is a huge advantage. That may involve reaching out to existing connections or investing in new relationships and showcasing your value’

The AI hiring black hole does not reject you because you lack talent. It rejects you because the system is running hot on spam and has no way to tell the difference.

That is the problem. Nobody built it on purpose. Nobody is fixing it on purpose either.

Iaroslav Belkin is the founder of Belkin Marketing, a strategic intelligence firm executing mission-critical engagements for AI, deep tech and frontier technology companies. Strategic partnerships, international expansion, executive positioning, VVIP events and access to the World Economic Forum in Davos.

See more breaking stories here.