What do you do when the job interviewer is an AI?

I build software that fills out job application forms. It never interviews anyone. But the two moments share a shape: a machine sits between you and someone’s hiring decision, and the honest question is what that machine is actually doing with what you give it, and how much choice you actually have in the room. I can answer that question for the tool I build. For whatever AI system is running your next interview, the answer isn’t obvious, but it is researchable, both how common this has gotten and what the law actually promises you.

How common this actually got in 2026

AI interviewers stopped being a novelty sometime in the last year. In a 2026 survey of 2,950 job seekers across the U.S., U.K., Germany, Australia, and Ireland, the recruiting platform Greenhouse found that 63% had now faced an AI interview, up 13 percentage points from six months earlier. The same survey found the experience mostly hasn’t been a good one: 38% of candidates said they had walked away from a hiring process specifically because it included an AI interview, and another 12% said they would. Seventy percent said no one clearly told them upfront that AI would be evaluating them, and 21% only found out once the interview had already started.

The rollout has been uneven enough to go viral for the wrong reasons. In one widely circulated case this year, NBC News reported that an Ohio State student interviewing for an entry-level position at StretchLab was greeted by an AI recruiting agent built by the startup Apriora. The bot glitched partway through and repeated the phrase “vertical bar Pilates” fourteen times in twenty-five seconds before the interview ended. The same reporting noted that AI recruiting agents from vendors like Apriora have already been used or tested by employers including the Boston Red Sox, Zillow, Chipotle, and McDonald’s.

What the law actually promises you, and it isn’t much

There is no federal law that requires an employer to tell you an AI is interviewing you, and none that guarantees you can ask for a person instead. What exists is a patchwork, state by state, and it’s worth knowing which state you’re actually applying in.

Illinois has two separate laws that touch this. The Artificial Intelligence Video Interview Act, in effect since 2020, requires an employer to notify you before your video interview, explain in general terms how the AI evaluates candidates, and get your consent before it analyzes your responses; on request, they have to destroy the video within 30 days. A newer amendment to the state’s Human Rights Act takes effect January 1, 2026, and goes further: it requires notice whenever AI plays a role in recruiting, hiring, or several other employment decisions, and explicitly bars employers from using AI in a way that discriminates, including using your zip code as a stand-in for race.

New York City’s Local Law 144 works differently. It requires employers using an “automated employment decision tool” to run an independent bias audit every year, post a summary of the results, and notify candidates that such a tool is in use. It does not require offering you a human alternative.

Colorado came closest to a right you could actually use, and then delayed it. The state’s rewritten AI Act, signed in May 2026, gives job applicants and other consumers the right to request “meaningful human review and reconsideration” of a high-stakes automated decision, though only “to the extent commercially reasonable.” The same bill pushed the effective date back to January 1, 2027, so as of this writing it isn’t enforceable yet. Outside these states, whether anyone tells you an AI is running your interview, or lets you skip it, is entirely up to the employer.

The one lever that doesn’t depend on your state

There’s one legal hook that isn’t tied to which state you live in: the Americans with Disabilities Act. The EEOC’s guidance on AI hiring tools is direct about this: an employer has to provide a reasonable accommodation when AI-based assessment might not rate a disabled applicant fairly, which can include a different interview format entirely. That’s a federal obligation, not a state one, and it applies whether or not your state has passed anything about AI at all. If a video-based AI interview genuinely doesn’t work for you, that is a specific, documented request an employer has to actually respond to. Asking “can I do this with a person instead” as a general preference is a request you’re always allowed to make. Whether an employer says yes to that one is up to them.

How to actually prepare for one

Most interview advice assumes a person is listening for tone, rapport, whether they’d want to sit next to you. An AI interviewer, or the model scoring the transcript of one, is doing something narrower: matching what you said against what the question was actually built to test for. That isn’t a worse form of evaluation. It’s a different one, and preparing for a conversation when you’re actually facing a structured test is where a lot of the bad interviews I’ve read about start.

Two things follow from that, and they’re both concrete. First, answer the question that was asked, completely, before you add color. A conversational answer that circles toward the point eventually reads as a non-answer to a system scoring for specific content. Second, know what the posting itself already told you the role needs, and make sure your answer touches those exact requirements early, not buried at the end. Vague enthusiasm doesn’t score. Specific, on-topic content does, whether the thing reading it is a person or software.

That second part is close to what I spend my actual working hours on, just on the application form instead of the interview. AppStride’s interview prep pulls the employer’s own required questions straight from the posting, along with the gaps its scoring already flagged when it compared that posting to your resume, so you’re not walking in with generic interview tips. You’re walking in with the specific things this employer already told you they’d ask about. Whether the interviewer across from you is a person or a screen, that part of the preparation doesn’t change.

None of this makes an AI interview feel less strange the first time you’re in one. But you’re not as powerless in it as it feels: ask whether a human option exists, and know your state’s actual rules instead of assuming there aren’t any. Then prepare for what the question is testing for, not who you imagine is listening on the other end. The format changed. What actually gets you through it didn’t.