Advanced program: Pro for the price of Plus for your first year. Limited places.Claim your spot
Recruitment

AI is screening your candidates. You're responsible for its decisions.

Applicant tracking systems rank CVs, chatbots run first interviews, and AI writes the job ads and the rejections. When a tool screens someone out unfairly, "the software did it" isn't a defence. The rules on this are now in force in several states, and the first cases have been decided.

What's changed

The rules and the numbers.

California: records for four years, and bias testing counts.

Since 1 October 2025, California's employment rules cover automated decision systems used in hiring. Employers must keep the data for four years, they're responsible for vendors acting on their behalf, and proof of anti-bias testing can be used in their defence.

Source: Mayer Brown, September 2025

Illinois: tell candidates.

Since 1 January 2026, Illinois employers must tell staff and applicants when AI is used in decisions such as hiring.

Source: Fisher Phillips on Illinois HB 3773

New York City: audit first.

Automated hiring tools for New York City jobs need an independent bias audit every year, a public summary and notice to candidates.

Source: NY State Comptroller audit of Local Law 144

Facts checked 16 November 2026
When it went wrong

Real cases.

iTutorGroup, 2023

The US Equal Employment Opportunity Commission's first AI hiring case. iTutorGroup's software automatically rejected women over 55 and men over 60. It paid $365,000 to settle.

A rule built into software is still your rule.

Source: Bloomberg Law, August 2023

Mobley v. Workday, ongoing

A job seeker over 40 says he was rejected from more than 100 jobs through employers using Workday's AI screening, often within hours. In May 2025 a federal court let applicants over 40 join as a group. The court has accepted that a software vendor could be treated as the employer's agent. Workday denies discrimination. A hearing on a wider class is set for March 2027.

The tool's maker and the business using it can both be on the hook.

Source: Lawyer Monthly, September 2026

Where it goes wrong in a business like yours

Where it goes wrong, and how we fix it.

AI screening candidates with no person checking.

How we fix it

Starter flags careers pages that mention AI screening without telling candidates. Plus adds a step where a person checks every AI-screened decision before a candidate is rejected.

Candidates not told AI is involved.

How we fix it

Plus gives you candidate notice wording for job ads and application pages, and keeps it current with Illinois, New York City and California's rules.

No record of how hiring decisions were made.

How we fix it

Plus sets out what to record when AI plays a part in a hiring decision, and how long to keep it, with training records to show your team knows.

Candidate data going into tools nobody approved.

CVs pasted into a free chatbot to "summarise".

How we fix it

Plus lists the tools approved for candidate data and trains your recruiters on it. Starter checks the tools on your website and what their terms say about AI.

Clients asking how you use AI to shortlist.

How we fix it

Pro gives you a ready answer pack for client questionnaires and a live badge they can check.

The free check

What it looks at.

Your website and careers pages, how they describe any AI in hiring, the tools they load, and your privacy notice. Score, top three gaps and a plan recommendation in about 30 seconds.

See what your careers pages tell candidates.

Your website, a few questions, and the three things to fix first.