The AI Hiring Paradox: When Both Sides of the Table Start Gaming Each Other

Recruiting has quietly become an arms race. Job seekers use AI to apply to more roles, faster, and to rewrite their resumes for every posting. Employers use AI to screen those applications faster, and to filter most of them back out. Both sides are optimizing for themselves, and the data now shows that neither side is winning.

The paradox, measured

Two separate 2024–2025 surveys, read together, describe the same collision from opposite sides. Among companies already using AI in hiring, 82% apply it specifically to resume screening 1. At the same time, 62% of employers say they now reject resumes the moment they read as generic, unpersonalized AI output 2. Screening AI and application AI are, in effect, built to defeat each other, and the survey data confirms both sides are doing exactly that at scale.

"We are trying to make the process better, but at the same time, the technology is making the process worse" is the kind of line recruiting leaders now say out loud in public forums, and Greenhouse's own CEO put a name to the pattern in mid-2026.

The 'doom loop'

Greenhouse CEO Daniel Chait described the dynamic bluntly in July 2026: employers and candidates are both using AI against each other, making the hiring market worse for everyone at once 3. It is not a metaphor so much as a description of the incentive structure, every improvement on one side triggers a countermeasure on the other.

Hiring managers who trust AI to decide faster/better

70%[5]

Job seekers who call AI hiring decisions fair

8%[5]

Applicants who trust AI to evaluate them fairly

26%[4]

Sources: [4], [5]

Trust is the first casualty, and it runs in opposite directions

Gartner's own July 2025 survey found that only 26% of job applicants trust AI to evaluate them fairly 4. Separately, Greenhouse's AI Trust Crisis Report: a survey of more than 4,100 job seekers, recruiters, and hiring managers across the US, UK, Ireland, and Germany, found that 70% of hiring managers trust AI to help them make faster, better decisions, while only 8% of job seekers call the result fair 5. That 62-point gap between what employers believe AI is doing and what candidates experience is, on its own, a bigger governance problem than any single screening statistic.

Where this is heading

Gartner projects that by 2028, one in four candidate profiles submitted worldwide will be fake 4. That is not a prediction about worse candidates, it is a prediction about an arms race that keeps escalating because neither side has a reason to stop unilaterally. A candidate who stops using AI to apply falls behind candidates who don't. An employer who stops using AI to screen drowns in volume competitors have already automated away.

A sharper filter is not the way out

better screening AI and better application AI are not opposing solutions, they are the same arms race, viewed from either side. Breaking the loop requires replacing the adversarial signal both sides are gaming with a shared one: a verified, explainable profile the candidate owns, scored against real criteria the employer can see and defend, rather than two black boxes trying to outguess each other. That is a structural fix, not a feature. That is the premise the next generation of hiring infrastructure is being built around.

This article synthesizes third-party market research current as of August 2026. Figures are attributed to their original surveys and should be reread against the linked primary sources before citation elsewhere.