Anyone else worried about AI resume tools inventing numbers you can't actually back up in an interview?

Anyone else worried about AI resume tools inventing numbers you can't actually back up in an interview?

There's a writeup where someone tested six major AI resume builders and found their bullet generators routinely invent fictional metrics, team sizes, tools, whatever sounds "more impactful," whenever the input you gave it doesn't have a hard number to begin with. You type something vague like "helped speed up our deploys" and it comes back with "reduced deployment time by 40% through CI/CD pipeline optimization" - specific-sounding, confident, and not something you actually said or measured. (atsverification.com/blog/ai-resume-builders-tested-2026 if anyone wants the details.)

That seems like a much bigger risk than the usual "does it sound robotic" concern people bring up about AI resumes. A recruiter skimming past generic phrasing costs you a callback. An interviewer asking "walk me through how you got that 40%" when you don't actually know puts you on the spot in the room, in real time, with someone deciding whether to hire you.

And it looks like this is landing on the hiring side too - a Robert Half survey from earlier this year found 65% of hiring managers say AI-enhanced resumes are making it harder to verify what candidates actually did, so interviewers seem to be getting more pointed about follow-up questions because of exactly this pattern.

Curious how people here handle this if they've used AI tools to help with resume writing. Do you go back and fact-check every specific claim before sending it out? Any near-misses where you almost sent something out with a number you couldn't have actually defended?