What Employers Actually Use AI For in Hiring

Job seekers are told that AI is deciding their fate and given no detail about what it does. The detail matters, because in several jurisdictions it comes with rights you can exercise. For an employer-side overview of computer activity records, see employee PC activity tracking.

This describes regulation as it stood in mid-2026 and is not legal advice. Requirements differ by where you are located and change quickly. Where something here matters to you, check the current position. For an independent public reference, consult EEOC artificial-intelligence resources.

What is actually in use

The tools that appear in modern recruiting stacks include resume screening engines that produce a fit score against a job description, video interview platforms that generate scores, algorithmic candidate matching inside applicant tracking systems, machine-learning personality and cognitive assessments, AI sourcing platforms, and promotion-recommendation tools.

Tools that merely transcribe, translate or capture data without scoring are treated differently — they are not producing a decision signal.

And a point worth knowing: the most common surprise for employers is finding these capabilities embedded in platforms they assumed were manual, because modern applicant tracking systems frequently include AI-powered ranking, recommendation and "candidate quality score" features that activate by default.

So the employer you applied to may be using scoring without having decided to.

What this does not mean

It does not mean your application was auto-rejected. Recruiters report that 92% reject manually or through knockout questions, with only 8% configuring content-based automatic rejection.

Ranking is not rejection. A tool that orders candidates changes who gets read first, which matters enormously given volume — but a person still decides. See what applicant tracking systems actually do.

The rights you may have

The part almost no job search advice mentions.

New York City

Local Law 144 covers automated employment decision tools used to screen candidates or evaluate employees for promotion. The statutory definition is any computational process derived from machine learning, statistical modelling, data analytics or artificial intelligence that issues a simplified output — a score, classification or recommendation — used to substantially assist or replace discretionary decision making.

Where it applies, the employer must have an independent bias audit conducted within the past year, publish a summary of that audit on its website, and give candidates at least ten business days' notice before the tool is used.

Applicants have the option to decline the use of these tools in their evaluation.

It applies based on where the candidate is located, not where the employer is headquartered — including for remote positions.

Penalties run from $500 to $1,500 per violation, with each day and each un-notified candidate counted separately.

Enforcement accelerated in 2026 after two years of a softer posture.

What this means for you if you are in New York City: you are entitled to notice, you can look for the published audit summary, and you can ask for an alternative process.

Illinois

The AI Video Interview Act has required consent and transparency for AI analysis of video interviews since 2020, and it prohibits sharing video content with third parties without consent.

A further law effective January 2026 is described as the only United States law giving candidates a private right of action against employers — meaning individuals can bring a claim directly rather than going through an agency first.

European Union

Under the EU AI Act, AI used for recruitment or selection is classified as high-risk — covering targeted job advertising, CV filtering, candidate evaluation, and tools deciding on promotion, termination or task allocation. High-risk obligations apply from 2 August 2026.

Employers deploying these tools must provide human oversight, monitor the system in use, keep automated logs for at least six months, and tell affected workers and candidates that the tool is in use.

Penalties reach €35 million or 7% of annual global revenue.

Elsewhere

Colorado's AI Act took effect in February 2026 and regulates high-risk AI systems including employment. Maryland requires consent for facial recognition in pre-employment interviews. California's Civil Rights Council has adopted automated decision system regulations.

Frameworks also exist or are developing in the UK, Canada and across the Asia-Pacific region, and they converge on bias auditing, candidate notice, impact assessments and human oversight.

The common principle: these laws apply based on where the candidate is located, not where the employer or the vendor sits.

What to do with this

Read the notice. If an employer tells you an automated tool will be used, that notice exists because it is required. It usually says what the tool assesses.

Look for the audit summary, where one is required to be published. It is on the employer's website and it tells you what was measured.

Ask what is being used, if you are unsure and it matters to you. "Will any automated screening or scoring be used in this process?" is a reasonable question for a recruiter.

Ask about an alternative where you have that right, or where you have a reason — a disability that a video assessment handles badly, for instance.

And weigh the cost. Exercising a right in a process you want to succeed in has a social cost, and that is a real consideration rather than one to dismiss. Knowing you have the option is worth something even where you choose not to use it.

Video interviews and recorded assessments

The area where the assessment is most opaque and where the regulation is most specific.

Some platforms score delivery, not only content — pace, word choice, sometimes facial and vocal characteristics. That last category has attracted the most legal attention and several vendors have moved away from it.

Consent requirements apply in several places, and where they do, you should be told what is being analysed.

If you have a condition that affects delivery — a speech difference, a neurological condition, anything affecting facial expression — an adjustment is usually available and asking for one is normal. See remote and recorded interviews.

What this changes about how you apply

Less than you might expect, and in a specific direction.

Do not optimise for a scoring system you cannot see. The configurations differ per employer, most of them are not doing content-based rejection, and the effort is better spent elsewhere. See why ATS optimisation tools sell you a problem.

Ranking rewards clarity, not keyword density. A record that plainly states what you did in the terms of the field ranks and reads better than one stuffed with terms.

The knockout questions still matter most. See screening questions.

And referrals bypass a great deal of this, because a referred application enters a different queue. See referrals.

The short version

Resume scoring, candidate ranking, video analysis and chatbots are all in use — and frequently switched on by default inside platforms nobody chose them in.

Ranking is not rejection. A person still decides in the large majority of cases.

In New York City you are entitled to notice, a published bias audit, and the option to decline. In Illinois, consent for AI video analysis. In the EU, notice and human oversight from August 2026.

These laws follow the candidate's location, not the employer's.

And none of it changes the advice: clarity, knockout questions, timing and referrals, in that order.