Yes, and for most companies this is already happening without anyone in HR or communications knowing what the answer sounds like. PerceptionX, an employer brand intelligence firm, surveyed 306 job seekers across the US, UK, Germany, Brazil, India, France and Japan in its 2026 report on AI and employer research and found that 96% of AI-using candidates had already asked a chatbot to research an employer, learn about a role, or help prepare for an interview. More strikingly, 82% said the AI's answer, or their own fact-checking of it, changed their mind about pursuing a job at a specific company. This is not a future risk. It is a current one, and it runs on the same mechanics that determine whether your company gets cited when a customer asks AI for a recommendation.
Source: PerceptionX, "How Job Seekers Use AI to Research Employers Report 2026," 306 respondents across 7 countries, linked in the article.
When do candidates actually turn to AI in the hiring process?
At nearly every stage, according to PerceptionX's data. Candidates reported using AI tools at 82% of the application-writing stage, 72% before applying at all, 62% during interview preparation, and 33% after receiving an offer, specifically to help decide whether to accept it. That last figure matters most to employer brand teams, because it means the research is not confined to the top of the funnel. A candidate who has already passed every interview and received a written offer may still ask ChatGPT what it knows about your company's layoffs, leadership turnover, or glassdoor sentiment before signing.
ChatGPT dominates as the tool of choice, used by 89% of candidates who research employers with AI, ahead of Google Gemini at 65% and Claude at 39%. The most common prompts are practical rather than exploratory: 70% use AI to prepare for an interview, 54% ask directly whether a company is worth pursuing, 54% ask what it's actually like to work there, and 40% use AI to surface employers they had not previously considered. That last number is the one that should concern a founder who assumes candidate discovery still runs entirely through job boards and referrals. Four in ten candidates are letting an AI model suggest companies to apply to that were not already on their list, based on whatever that model has learned about your company from the open web.
How much do candidates actually trust what AI tells them?
Not fully, but not little either. PerceptionX found that only 5% of candidates take an AI's answer about an employer at face value. Another 42% say they mostly trust it but verify occasionally, and 41% say they actively verify important claims before acting on them. The most common way they verify is unglamorous: 66% go check the company website directly, 66% run their own Google search, and 48% check LinkedIn. That pattern should sound familiar to anyone who has read how AI visibility works for customer-facing searches: the AI answer is rarely the final word, it is the first stop that shapes what the person looks for next, and what they look for next is still your own site, independent coverage, and your social presence.
The verification habit also means a wrong or outdated answer does not just slip past unnoticed. PerceptionX reports that 58% of candidates say they have personally caught an AI assistant giving inaccurate information about an employer, whether a stale fact, an old executive name, or a misreported layoff. Once a candidate catches that kind of error, the fix-it burden falls on the company's own public information to be current enough that the correction is obvious and immediate.
Why does this work the same way as customer-facing AI visibility?
Because it is the same underlying mechanism; only the question changes. Earned media coverage already carries disproportionate weight when an AI model is asked who to buy from in a given category, because a model trained on the open web treats independent reporting as more trustworthy than a company's own marketing copy. The same holds when the question shifts from "which company should I buy from" to "which company should I work for." A model answering a candidate's question about your culture, stability, or leadership is pulling from whatever has been written about you that it can access: news coverage, employee review platforms, your own careers page, LinkedIn posts from current and former staff, and any crisis coverage that is still indexed.
This is also why a company's online review record matters for AI citations beyond the customer-facing use case. A Glassdoor or Indeed profile with a thin, stale, or heavily negative set of reviews is exactly the kind of signal a model repeats back to a candidate who asks what it's like to work somewhere, in the same way a sparse Trustpilot profile limits whether a model will cite a brand to a shopper. Treating employee reviews as a communications asset, not an HR afterthought, is no longer optional once a meaningful share of candidates are reading an AI's summary of them before anyone from your company gets a chance to make a first impression directly.
What does this mean for a funding announcement or a new hire story?
It raises the stakes on getting the basic facts right and current. A candidate asking AI about a company that just closed a funding round, or just announced a new executive hire, is going to get an answer shaped by how that news was covered, if it was covered at all. A round that only appears in a press release on the company's own site gives a model one thin, self-interested source to work from. The same round covered by an actual outlet, even a trade publication, gives the model independent language to draw on when a candidate asks whether the company is stable, growing, or a reasonable bet to join. How a funding round gets announced is no longer purely an investor-relations or customer-acquisition question; it is also feeding the answer a prospective hire gets eighteen months from now when they ask an AI model whether the company is still around and still growing.
The same logic applies in reverse during a crisis. A company that handles a layoff, a data breach, or a leadership departure with a clear, factual public statement gives AI models something accurate and current to repeat. A company that says nothing publicly leaves the model to assemble an answer from whatever secondhand coverage, forum posts, or outdated articles it can find instead, which is rarely more flattering than the truth would have been. A basic crisis communication plan written for customers and investors already covers most of what a candidate-facing AI answer needs too.
What should a company actually check first?
Three things, in order of how fast they are likely to move the answer a candidate gets.
- Ask ChatGPT, Gemini and Claude directly what they know about your company as an employer, and note anything outdated, incomplete, or wrong. This costs nothing and takes ten minutes.
- Audit your Glassdoor, Indeed, and LinkedIn presence the way you would a customer review profile: is it current, is the volume high enough to look credible, and has anyone from the company responded to negative reviews with something other than silence.
- Check whether your most recent major news, a funding round, a leadership change, an expansion, is covered anywhere independent of your own website. If the only source is your own press release, that is also the only source a model asking about you as an employer has to work with.
None of this guarantees a specific answer from a specific model, and PerceptionX's own figures describe a single vendor's survey of 306 respondents, not an audited measurement of what AI models actually say about named employers. What the data does show clearly is that the behavior, asking AI before applying, before interviewing, and before accepting, is already common and already shaping decisions, which makes the underlying information worth getting right regardless of how any one model happens to answer today.
The direct answer, for anyone who read this far
Yes. A large majority of candidates who use AI tools at all are already using them to research employers, and a majority of those say the answer changed their mind. The fix is the same discipline that works for customer-facing AI visibility: keep your major news covered somewhere independent of your own site, keep your review platforms current, and check what the major AI assistants currently say about you as an employer before a candidate does it for you.
If you want help making sure your company's coverage, reviews and public record hold up when an AI assistant gets asked about you, get in touch and we'll walk through what's missing.
