AI in hiring: what you are entitled to know

This is information, not legal advice, and it is current as at August 2026. This area is moving quickly: one of the laws described below was repealed before it ever took effect, and another is in force but with its enforcement stayed by a federal court. If you are relying on any of this for a decision that matters, check the primary source, which is linked at the end.

The common picture is a machine reading your resume and deciding. That is rarely what happens.

What almost always happens is parsing and ranking. The system extracts your details into fields, indexes them, and helps a recruiter search and sort. A poorly parsed resume is harder to find in that search, which costs you quietly rather than producing a rejection.

What genuinely rejects automatically is a knockout question, and that is a rule a person configured: work authorisation, a required licence, a minimum years-of-experience gate answered yes or no. If your application disappeared instantly, that is the likelier explanation than a model reading your bullets, and it is covered in knockout questions.

We wrote about where the belief that most resumes are auto-rejected came from, because it has no study behind it: where the 75% number came from.

Local Law 144 covers automated employment decision tools used to screen candidates or evaluate employees for promotion in New York City.

If one is used on you, the employer must give clear and timely notice, must say which job qualifications and characteristics the tool assesses, and must tell you that you can request an alternative selection process or a reasonable accommodation. For a promotion decision about an existing employee, that notice has to come at least ten business days beforehand.

Employers also have to publish a summary of the most recent bias audit of the tool, clearly and conspicuously, on their website. That is a document you can go and read before you apply.

Worth knowing how well this is policed. On 2 December 2025 the New York State Comptroller published an audit concluding that the Department of Consumer and Worker Protection's enforcement has been ineffective. Three quarters of test calls to 311 about these tools were misrouted and never reached the department, and while the department's own survey of 32 companies found a single case of non-compliance, the Comptroller's auditors reviewing the same companies identified at least 17 potential violations. Expect the rule to be enforced more actively from here, and do not assume an employer is complying because nobody has stopped them.

Full detail on this rule.

HB 3773 amends the Illinois Human Rights Act and took effect on 1 January 2026.

It requires an employer to notify you when AI is used in recruitment, hiring, promotion and other employment decisions. It separately prohibits using a zip code as a proxy for a protected class, which is aimed at a specific and well documented way that a model reproduces discrimination without ever being told to.

One caveat that matters for what you can expect in practice. The Illinois Department of Human Rights proposed rules setting out the timing, wording and delivery of that notice, and then withdrew them. The statutory duty stands; the detail of how notice must be given is unsettled, so what you actually receive may vary between employers.

Full detail on this rule.

Since 1 January 2026, Ontario's Employment Standards Act requires an employer with 25 or more employees to disclose in any publicly advertised job posting whether artificial intelligence is used to screen, assess or select applicants.

This is the most useful of the three for a job seeker, because the disclosure arrives before you apply rather than after. If you are applying in Ontario to an employer of that size and the posting says nothing, that is itself informative.

Employers below 25 employees are outside the requirement.

Full detail on this rule.

This one is worth stating carefully, because a great deal of writing about it is out of date.

Colorado SB 24-205, widely described as the first comprehensive US AI act, was repealed before it ever came into force. Its implementation was postponed, xAI filed suit in April 2026 seeking to block it, a federal magistrate stayed enforcement later that month, and the governor signed a replacement, SB 26-189, on 14 May 2026.

The replacement takes effect on 1 January 2027 and drops the original's discrimination duty in favour of disclosure. Its enforcement is currently on hold pending the same federal litigation. So as at August 2026 there is no Colorado AI hiring law you can rely on, and anything telling you otherwise is describing a statute that was withdrawn.

Full detail on this rule.

These four were missing from the first version of this page because we had not checked them properly yet. They are here now, and two of them turn out not to be in force.

California has covered automated decision systems under its Fair Employment and Housing Act since October 2025, for employers with five or more staff. It is strong on bias testing and four year record keeping, and it gives you no notice right at all, which is the main way it differs from New York City.

The United Kingdom has the strongest rights on this page and the least well known, because nothing about them mentions employment. Since February 2026, a significant decision taken with no meaningful human involvement carries a right to information, to make representations, to human intervention and to contest it.

Australia is not in force yet. The new transparency obligation commences on 10 December 2026, and it puts the disclosure in an organisation’s privacy policy rather than in a notice to you.

The European Union is not in force yet either. The AI Act classifies CV screening and candidate ranking as high risk, and those obligations were postponed from August 2026 to 2 December 2027. Until then your protection in the EU comes from the GDPR.

In New York City, ask for the bias audit summary before you apply, and if a tool is used on you, ask for the alternative selection process the law says you may request.

In Illinois, ask whether AI was used in the decision, since you are entitled to be told.

In Ontario, read the posting. If the employer has 25 or more staff, the answer should already be there.

Everywhere, you can ask a recruiter directly whether an automated tool screened your application and what it looked at. None of these laws gives you a right to see your own score or ranking, so that is a request rather than an entitlement, and how it lands tells you something about the employer either way.

Less than the coverage suggests. The advice that follows from all of this is the same advice that follows from wanting a person to read you clearly: one column, contact details in the body rather than a header, standard section headings, a real text layer in the file, and the language of the job you are applying for rather than a thesaurus of it.

Write it so software can read it because that makes you findable. Do not write it for software, because a person still makes the decision.

What a parser actually does with the file, and which formatting choices genuinely cause trouble, is covered in what an ATS actually does with your resume.