Employers Are Making Job Candidates Jump Through Hoops to Prove They’re Real

There’s a new frontier in job interviews and it goes something like this: Remove your Zoom background and pan your camera around the room. Now, could you please wave your hand in front of your face?

The ubiquity of AI-assisted answering—not to mention the risk of hiring candidates who might be lying about where they live, or even working for North Korea—has companies upping their screening tactics. They want to ensure the people they’re hiring really know what they’re talking about and are who they say they are.

Heather Wilcox, who has worked as a quality engineer, is actively job hunting now. She says she has been subject to interviews conducted by AI in which software monitors where her eyes are looking to ensure she’s not glancing at an AI-generated script.

She’s also been told by interviewers that virtual video-call backgrounds are a no-no. But as a former hiring manager herself, she gets it.

“There is a lot of person-swapping and cheating that goes on,” she says. “AI has made hiring super difficult because everybody’s résumé is super perfect. Everybody has the right answer.”

While many companies want to see how adept job seekers are at using artificial intelligence and might encourage its use, there are certain questions they still want candidates to answer without help: how they think through a particular problem, specifics of their prior work experience or how they got through a challenging situation at a prior job, for example.

A September survey of 3,000 managers by Checkr, a verification platform, found that 59% of managers suspected job candidates had help from AI tools to misrepresent themselves during the hiring process.

“This is a reality for every company that hires remote,” says Benjamin Sesser, general manager of BrightHire, a Zoom company that makes software to conduct, record and analyze remote job interviews. “It’s very problematic.”

BrightHire added a cheating-detection tool to its AI-interviewing product earlier this year that can monitor certain browser signals: “It’s monitoring: Are you switching browser tabs? What are you doing on your machine?” says Sesser.

When candidates interview with the telecommunications and security company Somos, they’re asked to take down their Zoom background and pan their camera around the room. The goal is to spot clues that they’re not where they claim to be or if there’s another person—or computer setup—in the room feeding them answers.

Some candidates disconnect the call upon hearing the request.

“We were getting a lot of fraudulent candidates,” says Elissa Barrett, chief people officer of Somos, which is based in New Jersey but is a remote-first company.

Somos recently started checking candidate IP addresses. They had been catching candidates claiming to be in Florida, for example, but actually residing in Brazil. They’ve also started asking job seekers to wave their hand in front of their face: A bot posing as a human with an AI-generated face can’t typically complete the challenge.

“When something is off, they hold their hand to the side,” says Barrett.

Mike Hauschild is a recruiter with Quantum, which hires candidates for more than 200 venture-backed startups. He says job seekers are encouraged to use AI for things like take-home assignments or coding exercises, but there are other questions they need to answer on their own.

One client requires candidates to put their hands in the air while the interview prompt is being delivered to prove there’s no live typing going on. “They phrase it as, we’re going to do a little bit of a hands-on screen exercise,” he says. “It’s wacky.”

Eva Spexard, head of AI operations for a 30-person remote-first startup called Passion.io, says her company asks candidates to share their screen for certain parts of the interview. They have also moved to written answers for some questions, providing answers in real-time in a shared Google doc.

“You basically see if they’re typing or if they’re copy-pasting,” she says.

Spexard says take-home assignments became so riddled with AI slop that they’re becoming obsolete and live assignments are now the way to go. There were too many times where it was clear a chatbot had been used, including “copy-pasted replies that still had the Claude sentence, ‘Should I make that more concise?’” says Spexard.

At Nominal—which builds software for hardware engineering teams and has around 220 in-office employees—candidates aren’t hired until after they come for an in-person interview. Coding tests for engineers are conducted on site.

“We can fully control the cheating experience through the on-site interview,” says Bryce Strauss, Nominal’s CEO. “We have a laptop that we provide them.”

In the past few years, Strauss says he’s noticed a widening gap in people’s abilities to think through problems: He suspects some are relying on AI so much in their day-to-day work that it’s affecting their ability to think on their own. When AI is being used to answer a question, a telltale sign is the answer ends up being mechanical and far less interesting.

“Tangents, false starts, weird connections you didn’t expect to make mid-thought tend to get cleaned up when AI is involved,” he says. “The way to stand out in this moment is to be fundamentally human.”