Which AI is screening your job application? Part 2: Your rights and your move

This is Part 2 of a two-part series. Part 1 walked through the machines that read your application first: resume grading, LinkedIn and Indeed agents, chatbot interviews and scored video interviews. This part is about leverage: what the law makes employers tell you, what you can do when you suspect the software got it wrong, and how to use AI on your side.

The theme running through all of it is accountability. Every tool in Part 1 makes a decision, or shapes one, about a person who never agreed to be evaluated by it. Someone has to own that decision, and the law is slowly deciding who.

Do employers have to tell you AI is screening you?

It depends on where the job is: New York City and Illinois require notice for specific tools, California regulates automated decision systems in employment, and in most of the country no law requires an employer to say anything. New York City’s Local Law 144, enforced since July 5, 2023, bars employers from using an automated employment decision tool unless it has an independent bias audit every year, the results are posted publicly, and candidates get notice at least 10 business days before it is used. Candidates can also request an alternative selection process or an accommodation.

Illinois’s Artificial Intelligence Video Interview Act, in effect since January 1, 2020, requires notice, an explanation, consent and deletion within 30 days of a request before AI analyzes a video interview. Since January 1, 2026, an amendment to the Illinois Human Rights Act also requires employers to notify applicants and employees when AI is used in hiring and other employment decisions. California’s rules on automated decision systems, covered on my AI page, apply existing anti-discrimination law to these tools. Interviews can be recorded too, see AI notetakers in meetings.

My position: New York City’s law should be the national floor, and California should require the same notice by name. If an employer can buy software that grades me, I should be able to read the audit of that software. Notice costs an employer one paragraph on a careers page. The objection was never cost. It is that notice invites questions, and questions are exactly the point.

What can you do if you think AI screened you out unfairly?

Ask the employer which tool evaluated you and whether a person reviewed your application, keep records of every step, and if the pattern points to discrimination, file a complaint with California’s Civil Rights Department or the federal EEOC before the deadline. Under California’s Fair Employment and Housing Act you generally have three years to file with the Civil Rights Department; the federal deadline is usually 300 days in California. Save the posting, your confirmation email and the time of any rejection. A rejection that arrives minutes after you applied is worth keeping.

The case to watch is Mobley v. Workday. Derek Mobley, a Black man over 40 with a disability, applied to more than 100 jobs at employers using Workday’s screening and was rejected every time. In May 2025, a federal court in San Francisco let his age discrimination claim proceed as a nationwide collective action for applicants 40 and older. In July 2025, the court held that the collective includes applicants scored, ranked or screened by HiredScore, the A, B and C grading tool from Part 1, and Workday had to identify the employers using it. The case is ongoing, and Workday denies the claims.

Here is why it matters. For years, employers pointed at the vendor and vendors pointed at the employer, and the applicant fell in between. Mobley tests whether the company that builds the grade can be held to answer for it. I think it should be. Whoever profits from a hiring decision owns the hiring decision. And because the federal EEOC has since deprioritized disparate impact cases, state agencies like California’s Civil Rights Department now carry more of that weight for applicants here.

Should you use ChatGPT, Claude, Gemini or Copilot to write your resume?

Yes, as a drafting and checking tool, but write the final words yourself and never use hidden text to game an AI screen. Greenhouse’s 2025 survey found that 41% of U.S. job seekers say they have used prompt injection, white text or hidden instructions meant to trick an AI filter, while 65% of U.S. hiring managers say they have caught applicants using AI deceptively.

I understand the temptation. When the screen feels rigged, rigging it back feels fair. It is still a bad trade. Prompt injection is cheating, and worse, it is a losing bet: it works only until someone looks, and the moment someone looks, you have handed them a reason to reject you that has nothing to do with your ability. Every argument in this series about holding employers accountable depends on applicants who can say they played it straight.

The quieter cost is sameness. When 244 people apply and many ran their resumes through the same few assistants, the bullets start to read alike and a grader has nothing to separate them. AI writes toward the average, and I explain why that costs people from neighborhoods the data underrepresents in the anomaly trap. The job you held through school, the family you support, the numbers you moved: those are what a model sands off first, and they are your edge. Pick the assistant by the task:

Assistant	Best use in a job search
ChatGPT (OpenAI)	Tailoring resume bullets to one posting and running mock interviews out loud
Claude (Anthropic)	Comparing a long posting against your full resume and listing what is missing
Gemini (Google)	Drafting inside Gmail and Google Docs
Microsoft Copilot	Drafting inside Word and Outlook
Perplexity	Researching an employer with cited sources you can check

Three rules apply to all of them. Give it your real facts, or it will invent generic ones. Check every number and name it produces. And never paste a Social Security number, your date of birth or another person’s information into any of them.

How do you research an employer with AI search?

Ask an AI search tool that cites its sources, such as Perplexity, ChatGPT search, Google’s AI Overviews and AI Mode, or Microsoft Copilot, about the employer’s recent news, layoffs, lawsuits and pay, then open the sources before you rely on the answer. These tools summarize fast and sometimes summarize wrong, including blending two companies with similar names. The links are the research; the summary is the table of contents. Search the employer’s name with the words “lawsuit,” “layoffs” and “Glassdoor” at minimum.

Then turn the machine around and search yourself. Recruiters use these tools too, and an AI answer about your name pulls from your LinkedIn profile, your public posts and anything else it can find. If the answer is thin, outdated or about someone else, fix your public profile, starting with how recruiters search LinkedIn. It is the same fix that helps you with the agents in Part 1, and it is the one piece of the chain you fully control.

What should employers ask before buying an AI hiring tool?

Ask for the vendor’s most recent bias audit, how applicant data is secured and deleted, whether a person reviews every rejection, and how candidates are told the tool is in use. Then add the question the McHire breach taught: who at the vendor can log in to applicant data? Require multifactor authentication on every administrator account, a retention schedule that deletes old records, and written notice of any exposure.

The rule I give clients is simpler than any checklist: if you cannot explain why the tool graded a candidate a C, you cannot defend that grade, so do not buy the tool. Mobley shows where unexplained grades end up. And train the recruiters who read the grades to question them rather than sort by them; teaching judgment instead of the interface is the approach I lay out in how to train your team on AI. Adopt New York City’s audit and notice standard even where you are not required to, and your applicants will see a company that has nothing to hide. My AI page covers the liability side in full.

If you are applying and want a plan built around how these systems actually read you, that is what the Career Blueprint is for. If you missed the machinery, start with Part 1.

McKinley holds a J.D. but is not a licensed attorney. Articles here are general information, not legal advice. For your specific situation, talk with an employment attorney.

Data current as of September 2026. Sources are linked where each figure appears.