Written assessments have been a staple of hiring for roles that require strong communication, analytical thinking, or domain expertise. The assumption underpinning them - that the submitted work reflects the candidate's own capabilities - has been quietly undermined. AI tools capable of producing polished, professional writing are now widely available and trivially easy to use. The question for talent teams is not whether candidates use them, but how to maintain the signal value of written assessments in an environment where they do.
The Scale of the Problem
Survey data from 2025 consistently showed that a substantial majority of job seekers had used AI assistance in application materials, with a significant proportion using it to generate cover letters or writing samples entirely. Among candidates for roles that involve writing as a core competency - marketing, communications, content, consulting, legal - the incentive to use AI is particularly high because the stakes of the written assessment are higher.
The practical effect is that recruiters increasingly cannot rely on written submissions as reliable proxies for a candidate's actual writing ability. The downstream consequence is hiring decisions made on false premises - and the discovery, often after onboarding, that a candidate's on-the-job writing does not match the quality of their application materials.
Why Simple Detection Is Not Enough
Running cover letters and writing samples through a detection tool seems like an obvious solution, but it carries two significant risks. First, false positives: candidates with clean, structured writing styles - often the strongest communicators - can score high on AI detection metrics and be incorrectly flagged. Rejecting qualified candidates on the basis of a probabilistic score is both a legal exposure and a talent acquisition failure. Second, false negatives: sophisticated candidates know how to produce AI-assisted content that evades common detectors, meaning the candidates most likely to attempt deception are also the most likely to succeed.
Shifting Toward Verified Assessment Workflows
A more durable approach is to require that writing assessments be completed through a structured authorship workflow, rather than analyzing submitted artifacts after the fact. When candidates complete assessments in an environment that captures provenance signals during composition, the resulting submission comes with contextual evidence that is far more reliable than post-hoc detection alone.
This approach also reduces legal risk. Rather than accusing a candidate of using AI based on a probabilistic score, organizations can establish a consistent policy requiring all assessments to be submitted through a verified workflow - a content-neutral requirement that applies equally to all candidates.
What to Look for in a Verification Platform
Talent teams evaluating content verification tools should look for platforms that provide structured verification reports rather than bare scores, surface confidence levels and flagged uncertainty, and support a documented review process for borderline cases. Platforms like AuthProof are designed to support this kind of structured, auditable verification workflow - providing organizations with the evidence they need to make defensible decisions rather than relying on opaque algorithmic outputs.