When the job offer is the trap: generative AI turns the job hunt into a minefield

The Doherty Staffing Solutions episode acts as an uncomfortable mirror. The Minneapolis agency had not sent out a single recruiting message, yet it began receiving calls from candidates responding to job offers bearing its logo, accompanied by impeccable photographs of a certain 'Emma Smith'. A forensic analysis confirmed what had already been suspected: those images had been fabricated with artificial intelligence. The goal was not to hire anyone, but to capture Social Security numbers and banking details in exchange for nonexistent jobs.

What matters here is not that scams exist —they always have— but the qualitative leap introduced by generative AI. Professor Manjeet Rege, of the University of St. Thomas, sums it up precisely: the cost of producing a convincing hoax has collapsed, and building a credible website is now a matter of minutes. The phrase that best captures the change is his own: 'don't believe it even if you see it'. The old advice of trusting visual cues —blurry logos, spelling mistakes— is running out of ammunition, because those flaws are disappearing at breakneck speed.

The figures outline the scale of the problem without any need for drama. According to the FTC, impersonations cost 3.5 billion dollars last year, nearly triple the figure of five years ago; the FBI logged close to 22,000 reports of AI-enabled fraud, with 893 million in losses. And Deloitte projects that deepfakes could push the annual cost of consumer fraud in the United States to 40 billion by 2027. These projections should be read with caution —they are estimates, not certainties— but the direction of the trend is unmistakable.

There is a nuance worth underscoring, because it shifts the focus away from the usual cliché. Researcher Marti DeLiema, of the University of Minnesota, introduces the concept of 'unmet need': we don't fall into the trap merely because of age or technical ignorance, but because a real urgency —finding a job— lowers our defenses. This explains why the labor market is such an effective vector: it combines economic pressure with the everyday use of digital platforms. The personalization described by Billy Doherty, the firm's president, tailoring messages to each candidate's interests, exploits exactly that emotional vulnerability.

The constructive takeaway is that defense can also be industrialized. If AI makes the attack cheaper, it also makes verification cheaper: official confirmation channels, sender authentication, digital literacy geared toward processes —does this company ask for sensitive data by SMS?— rather than toward visual details. For staffing agencies, as the industry's trade body warns, more than money is at stake: the reputation of a legitimate channel on which hundreds of thousands of workers depend. AI-powered fraud is a serious problem, but it is above all a problem of design and habits, and both are correctable.