Quoting Tom MacWright

The sharper point MacWright is making, which the summary gestures at but doesn't fully land, is that the problem is not fraud in the traditional sense. Candidates using LLMs to generate materials are not necessarily misrepresenting credentials; they are producing technically accurate but identity-stripped outputs, which breaks hiring in a subtler and harder-to-litigate way.

This is largely disconnected from recent activity in our archive, as we have no prior coverage to anchor it to. It belongs, though, to a cluster of emerging stories about second-order labor market effects of LLM adoption, distinct from the first-order productivity framing that dominates most AI coverage. The relevant adjacent territory is how evaluation infrastructure (hiring, peer review, credentialing) degrades when the signal it was designed to read gets cheapened at scale. That is a different problem than capability or safety, and it is getting less attention than it deserves.

Watch whether major applicant tracking system vendors (Greenhouse, Lever, Ashby) ship explicit LLM-detection or provenance features within the next two quarters. If they do, it confirms the problem has crossed from blog-post concern to enterprise procurement priority.