There's a particular kind of dishonesty that doesn't feel like lying while you're doing it. You open your LinkedIn profile, click into a job you left in 2016, and add a line about "leveraging AI-driven insights to optimize team workflows." Never mind that the tools you're describing didn't exist yet. It doesn't feel like fabrication. It feels like translation — updating old language into the dialect employers currently speak.
A new working paper out of Stanford gives this behavior a name: time traveling. Researchers combing through years of archived LinkedIn snapshots found that roughly one in five long-time users have gone back and edited descriptions of jobs they no longer hold, often years after the fact, often to insert AI-related buzzwords that would have been anachronistic at the time. The median gap between leaving a job and rewriting its description was 4.5 years. Some edits reached all the way back to 2012 — a full decade before ChatGPT existed.
That detail is the one that should give everyone pause. This isn't a story about people slightly oversharpening what they already did. It's a story about people quietly installing capabilities into their own past.
The logic makes uncomfortable sense
None of this is happening because people suddenly became more willing to lie. It's happening because the market taught them that the alternative — telling the truth about a resume that doesn't check every keyword box — often means never getting past the applicant tracking system at all. When nearly three-quarters of employers say AI skills are now a requirement or a strong advantage, and job postings mentioning "AI" have tripled since 2022, the incentive isn't subtle. Show the right words or don't get seen.
That's the uncomfortable center of the study: the people time-traveling aren't cartoon villains. They're mostly acting rationally inside a system that rewards keyword density over truth density. The self-described "AI product manager" — the single biggest offender category, with about 70% making retroactive edits — probably isn't inventing a career from nothing. More likely, someone who managed a product with any tangential connection to AI, or none at all, is reaching backward to make yesterday's job title match today's hiring algorithm.
The part that should worry hiring managers more than job seekers
It's tempting to read this as a story about individual dishonesty. I think it's actually a story about what happens when a signal gets gamed faster than anyone can fix it.
LinkedIn profiles, resumes, keyword-matched job titles — these were never perfect signals of ability. But they used to at least be anchored to something: a job you had, at a time you had it, described in the language of that time. Time traveling severs that anchor. If a meaningful fraction of professional profiles are retroactively rewritten to match whatever skill is fashionable this quarter, then the entire signal degrades for everyone, including the honest majority who don't touch their old job descriptions.
And the fact that this behavior spikes right before people change jobs makes clear it isn't an accident of memory or a good-faith cleanup. It's targeted. It's strategic. It's an arms race, as one of the study's authors bluntly put it — and arms races don't resolve themselves. They escalate until something outside the system forces a correction: better verification, more skeptical recruiters, or eventually a market that stops trusting self-reported credentials altogether.
What doesn't get said out loud
The study is careful — appropriately so — to note that not all of this is malicious. If you weren't active on LinkedIn during an old job, updating that entry now technically counts as "time travel" even if you're just filling in a job you always had. Removing a duplicate word counts too. Some of this is noise.
But noise doesn't explain a decade-old product manager role suddenly claiming machine learning expertise that didn't exist as a job requirement in 2012. That's not tidying up. That's a bet — a bet that no one on the other side of the hiring process is checking closely enough to notice, and that the benefit of clearing an algorithmic filter outweighs the risk of getting caught in a conversation you can't actually back up.
It's a bet a lot of people seem willing to take. Whether it pays off probably depends less on how convincing the LinkedIn edit is, and more on how good the follow-up questions are in the room where it actually matters.