What an Employer Can and Can't See About Your AI Use

What Can They See From Your Own Account?

Nothing at all, structurally. A personal ChatGPT or other AI account, opened on your own device and never shared with anyone, has no channel back to an employer. They see the output you choose to send them, a resume, a cover letter, a take-home file, and nothing about how it got made unless you tell them or it's visible in the writing itself.

This holds regardless of how much or how little you used the tool. There's no general employer visibility into a candidate's personal AI accounts, no matter what a rumor from a friend of a friend implies, and no legitimate hiring process claims otherwise. The boundary is the account and the device, not the fact that AI was involved at all.

The confusion usually comes from somewhere else: a story about a company that somehow knew, which on inspection almost always turns out to be something else entirely, a writing sample that read like everyone else's, a follow-up question the candidate couldn't answer, a detector that flagged something and was wrong. None of those is the employer reaching into a private account. They're downstream consequences of the work itself, which is a different and more useful thing to think about than a surveillance story with no mechanism behind it.

It's also worth separating a personal account from a personal device used for work purposes. If you ever logged into a company's own systems, email, an applicant portal, a scheduling tool, on that same device, whatever visibility exists runs through that separate relationship, not through your AI account. The device itself isn't the boundary; the account and who controls the environment are, and that distinction is what the rest of this article is actually organized around.

What Changes Inside an Employer's Own Environment?

The moment you're inside something the employer built or chose, a proctored assessment, a sandbox environment, a one-way video platform, the picture is different, and it should have been disclosed to you before you ever started the session. What gets captured there depends entirely on that specific tool, not on a general rule that applies the same way everywhere.

One of the largest one-way interview vendors publishes exactly what its own AI scoring does and doesn't use: only the transcribed words of your answer, explicitly not facial expression, body language, background, or tone of voice ¹. That's a narrower claim than most candidates assume going in, and it's worth reading literally: the recording itself still exists and a recruiter can still watch it, even though the score that vendor's AI produces is built only from the transcript.

The same vendor's published process lets a candidate opt out of AI scoring specifically, with a human rating the same answers instead using the same criteria, and opting out is stated not to remove you from consideration on its own ². That's one vendor's documented behavior, not a universal guarantee: consent screens are configured by the employer deploying the platform, so what you're actually shown can differ from the document's example, and nothing here says every other platform in the industry makes the same choice available.

Illinois goes further and makes this a legal requirement rather than a vendor choice: since January 1, 2020, for an Illinois-based role, an employer using AI to analyze a recorded video interview must notify you beforehand, explain in general terms how the AI works, and get your consent before any analysis happens ³. Illinois isn't alone: New York City has required notice at least 10 business days before an automated employment decision tool is used on a NYC candidate since the rule took effect on January 1, 2023 ⁴. Most places have no equivalent statute, which is exactly why asking directly still matters even where no law forces the answer.

Ask What's Captured Before the Round Starts

In a well-run process, visibility is disclosed rather than hidden, and it's worth asking about directly rather than guessing, because guessing at what a tool captures is where both the paranoia and the false confidence start. Before a proctored or AI-scored round, it's reasonable to ask what's recorded, whether it's a tab or the full desktop, and how long any of it is kept.

A short list of what's worth asking specifically before you agree to anything: whether the capture is the assignment tab or the entire desktop, whether audio is recorded on its own separate toggle, whether any AI-scoring output is retained after the hiring decision is made, and who besides the immediate hiring team can actually see the recording. None of these are unusual questions, and a recruiter who answers them plainly is telling you something useful about the process behind the tool.

Employers running interviews are increasingly told candidates might record the session themselves and run it through AI afterward, which is the mirror image of this same visibility question from their side of the table, worth reading if you're weighing whether to keep your own copy of a session for your own records. The short version from that side: your own right to record depends on your state's consent rules, the same way an employer's right to record you does, and neither side automatically wins that question everywhere.

Separately, if what's actually worrying you is whether AI use itself gets noticed rather than whether the session is recorded, that's a different mechanism with its own honest answer: detection of AI-written text doesn't work reliably, which is a different fact from whether a specific tool captured your screen. Conflating the two is where most of the anxiety in this territory actually comes from, since one is a question about a recording device and the other is a claim about a piece of software's accuracy, and they deserve separate answers rather than one shared worry.

Read the evidence >