Candidates can use LLMs to generate real-time answers, rely on scripted prompts, or even employ hidden “whispering” helpers during tests. This undermines fairness, skews evaluations, and puts companies at risk of hiring candidates who can’t actually do the job.