If Software Reads the Resumes, Somebody Has to Say So

A four-person shop posts a job, gets a hundred and forty applications in a week, and does what anyone would do: pastes them into an AI assistant and asks it to rank the ten worth calling. The hiring gets done in an afternoon instead of a fortnight.

That is a real advantage, and it now comes with a condition that most small employers have not heard about. In Illinois, since the first of January, using AI to influence a hiring decision without telling applicants is itself a violation, independent of whether the software did anything unfair.

Sponsored by Stocks To Trade

Last month, Tesla ended production of the Model S.

Fourteen years. Gone.

They gutted the entire Fremont factory line and converted it to build something else…

Something Elon Musk says will eventually be worth more than the entire car business.

Production on the new line starts this summer.

Nvidia's CEO called what's coming 'the ChatGPT moment for robotics' at CES 2026.

And just like AI, the biggest gains won't go to the investors who buy the obvious name.

They'll go to the people who find the small companies no one is talking about yet.

A trader who turned $2,000 into over $2 million before 30 has identified five of them.

Two are trading under $4. He put the full breakdown in a short presentation.

See what they built instead →

Look Behind the Curtain

Illinois amended its Human Rights Act through House Bill 3773, effective January 1, 2026. Two things are prohibited. The first is using AI in hiring or promotion in a way that discriminates on a protected characteristic, which was already unlawful when a human did it. The second is newer and catches people out: failing to give notice that AI is being used at all.

The notice is specific. It has to name the product and its developer, say which employment decisions it affects, describe what data it processes and what it does, list the kinds of postings it applies to, give a contact for questions, and explain how to request an accommodation. It goes to applicants and, annually, to current employees, in plain language.

The scope is broader than most owners assume. It applies to employers under the state's Human Rights Act and to their agents, which includes recruiters and any third-party tool acting on your behalf. And the covered uses are ordinary ones: resume screening, targeted job advertising, computer-based assessments, and analysis of an interview recording.

Illinois is not alone. New York City has required bias audits and candidate notice for automated hiring tools since 2023, Colorado has passed a broader AI act reaching consequential decisions including employment, and California has adopted anti-discrimination rules aimed at automated decision systems. The details differ; the direction does not.

The mechanism behind the discrimination concern is worth understanding rather than fearing. These systems rank by pattern, and patterns in past hiring carry the preferences of past hiring. A model that learns your last twenty hires learns your last twenty biases, and applies them consistently and at speed.

The Capability Multiplier

The practical move is not to stop using AI in hiring. It is to use it for the part that is safe and keep the judgment where it belongs.

Drafting is safe and genuinely valuable. A well-written job post that states the pay range, the actual daily work and the real requirements gets better applicants, and most small employers write bad posts because writing is not their job. Preparing structured interview questions tied to the work is safe, and it improves hiring quality more than any screening tool.

Summarizing is the useful middle. Asking an assistant to pull out each candidate's relevant experience against a list you wrote, without ranking or scoring them, keeps you reading every application while removing the tedium. That is a different activity from asking it who the best ten are, and the difference matters both legally and practically.

Ranking and scoring is where the exposure sits. If you do it, the notice requirement applies where you operate, and you should be able to say what the tool weighed. A cheap discipline that solves most of this: write your criteria down before you look at anyone, and use those as the only thing the assistant compares against.

The quiet advantage in all of this is that a small employer can actually read every application. A large one cannot, which is why they automate the decision. You can automate the reading and keep the decision.

The Sovereign Action

An hour of setup covers a year of hiring.

- Check the rules where you employ people. Illinois, New York City, Colorado and California all have requirements now, and more states are following.

- If any tool influences screening or promotion, publish the notice: product name and developer, what it affects, what data it uses, who to contact, how to ask for an accommodation.

- Write your criteria before reading applications, and have the assistant compare against those rather than asking it to pick winners.

- Keep the shortlist decision human, and keep a short record of why each person advanced.

- Check whether your job board or recruiter is running automated screening on your behalf. Their tool is still your obligation.