Who’s really hiring you — HR or cyber security?

What would you think if you were asked in a job interview to hold up three fingers in front of your face? Perhaps you’re doing so badly that they’re checking you’re not concussed? More likely the recruiter is trying to check that you’re not an AI deepfake.

Employers are trying all sorts of things to combat fraudulent AI use in the recruitment process. There are jobseekers who use chatbots to bluff their way into jobs they’re not capable of, or pay “interview mules” to complete assessments for them. More dangerously, there are hackers who try to get hired with false identities in order to access internal systems, and even remote IT workers who turn out to be based in North Korea. The research company Gartner has predicted that one in four candidate profiles worldwide will be fake by 2028.

All this is leading to a culture clash inside companies. The warm and fuzzy realms of HR and recruitment are having to learn to work with the suspicious minds of the cyber security world.

One example: the HR platform Deel acquired AI cyber security company Clarity in August in order to “embed continuous identity security across the entire hiring and workforce lifecycle”.

Neev Wilf, who was at Clarity and is now head of product at Deel Security, tells me that “bringing HR and security to the same room is tough” because they “have different interests”. Security teams usually want “every protection, every feature, all the way” but “at the same time, you have [recruiters] who say ‘this enhances friction for the candidate’”.

What sort of security checks are we talking about here? Recruitment platform Greenhouse has introduced a system to allow employers to prompt job seekers for ID verification. “Some do it at the top of the funnel, in the beginning of the process, and some do it further down,” Daniel Chait, Greenhouse’s chief executive, tells me.

Clarity, meanwhile, says it uses AI to examine “facial movements, lip-syncing, eye motions, voice patterns” and other indicators in video interviews, and cross-checks indicators such as “behavioral, identity, and media patterns” to spot signals of deception.

It sounds like a lot of personal data to be hoovered up in a process that might not even result in a job offer. There is surely the potential, too, for “false positives”: for legitimate candidates to be flagged as suspicious and thus denied the chance of getting hired. If a lot of employers start to use the same handful of systems, some unlucky souls might find themselves blocked again and again.

The good news is that the EU’s penchant for regulation has helped pre-empt some of these potential pitfalls. Both GDPR and the EU AI Act give people protection against being subject to unexplained automated decision-making, for example, especially in high-stakes areas such as recruitment.

Both Wilf and Chait tell me their systems have been carefully designed to meet EU regulations, and do not lead to automated decisions but only flag information for human review.

That said, these new security features and identity checks feel like a depressing consequence of the AI arms race in recruitment. Already, candidates complain they feel dehumanised by employers’ use of the technology, while employers say they can’t cope with the volume of AI-assisted applications.

According to Gartner’s regular surveys of job seekers, the number of people who had experienced an AI-led interview doubled from 9 per cent in the third quarter of last year to 18 per cent in the second quarter of this year. But only 30 per cent said they felt comfortable with the experience, only 20 per cent felt they were able to get their own questions about the job answered and only 17 per cent preferred the experience to a human interviewer.

Have we reached the point where employers throw up their hands and go back to a much smaller pool of handwritten applications and in-person interviews?

It’s an appealing thought, but it’s worth remembering that way of doing things was hardly perfect either. Indeed, perhaps the worst-case scenario would be a future that contains a bit of both. A return to more offline, human-to-human recruitment decisions for those who already have personal connections who can vouch for them (or those young people whose parents have connections). And for job seekers without that reserve of social capital? An interview into a webcam with three fingers pressed to your face.

sarah.oconnor@ft.com