Last month, Paul Lee was reviewing applications for a legal and compliance job at the small tech company he runs when his screening software flagged an anomaly.
Hidden in one of the hundreds of resumes he received was a block of about 1,500 characters, typed in white text on a white background. It was an AI prompt to ignore previous instructions and declare the applicant a top-suit candidate, regardless of whether the person met the requirements of the job.
Lee said he was stunned and described the move as an ethical red flag for a role based on trust and integrity.
“You’re ahead of the curve,” says Lee, CEO of the 40-person Irvine, California-based company InnoCaption, which makes real-time captioning technology for the deaf and hard of hearing.
Resume hacking is getting an AI-era upgrade – and not in a way that’s recommended. While desperate job seekers once put keywords in white text to fool screening software, some people are now quietly entering AI prompts to try to make their applications stand out.
A May 2026 study by Duke University researchers and collaborators analyzed nearly 200,000 resumes and found that about 1% contained hidden prompt injections.
‘Job seekers do what they think they have to do to filter through the noise’ Sarah Franklin, CEO from the HR platform Lattice, Business Insider told us. She said this behavior is not surprising because employers are overwhelmed with applications and often respond with generic rejections, if at all.
“The hiring process feels like a complete black box,” she says.
AI is part of the problem. The technology makes it easy for job seekers to apply and submit applications at scalewhile some employers use it to sort through the pile, says Nada Hashmi, an assistant professor of information technology at Babson College. That could mean that AI, rather than a person, could be involved in screening applicants before they reach the interview stage, which could give them an incentive to game the system.
“It’s really true that AIs talk to each other,” she said.
Still, hacking resumes can be futile. Recruiters and hiring managers often review applications in batches and move on once they’ve found enough promising candidatesleaving subsequent submissions unchecked, said Nathalia Aryani, corporate human resources director at property management company Terra Vista.
“The problem is the volume,” she said.
Even worse, getting noticed for this tactic can have consequences. Those caught can be seen as dishonest, and some employers are now actively looking for leads as word spreads.
At Zurich-based flight search startup Zerolook, co-founder and CEO Simone Lini last month uploaded about 20% of the roughly 350 resumes he received for two tech roles to an AI tool. He had heard of applicants hiding rapid injections and asked for any examples to be flagged.
One resume, Lini discovered, did indeed include instructions to ignore previous instructions and prioritize their candidacy, saying “their livelihood depends on you doing that,” he recalled.
Lini, whose company is hiring after raising $1.9 million in venture capital, said while the applicant had relevant experience, the ploy was a turnoff. He added that he disagreed with some commenters on a LinkedIn post he made about the issue who suggested it showed AI knowledge.
“It’s about trying to sneak in and get to the top,” he said, adding that inserting the prompt wasn’t difficult. He said he blocked the person from applying again.
