AI screening tools aren't fringe anymore. A 2025 Resume.org survey cited by Business Insider found that 57% of companies were already using AI somewhere in hiring workflows.
Among those employers, 79% said they use AI to review resumes. And 74% said an AI system could reject a candidate without human review.
That does not mean every AI hiring tool is bad. It means the first question cannot just be "does this save time?" It has to be "what would this miss?"
This week: what AI hiring tools actually do, where they can go sideways, and three questions to ask before you trust one.
───
🗞️ THIS WEEK IN AI
1. Anthropic's Fable 5 shutdown showed how fast AI access can change
On June 9, Anthropic released Claude Fable 5. A few days later, The Verge reported that the U.S. government issued an export-control directive restricting access to Fable 5 and Mythos 5 by foreign nationals. Because Anthropic said it could not reliably separate users that way in real time, the company disabled the models globally while it worked on a resolution.
Why this matters: If your team depends on one AI tool for real work, access can change faster than your workflow can adapt. Keep a backup tool or manual path for anything you cannot afford to pause.
2. Researchers found AI resume screeners can favor AI-written resumes
A recent paper on AI self-preferencing in hiring tested more than 2,200 resumes across 24 occupations. The researchers found that AI evaluators were more likely to shortlist resumes written by the same AI model used to judge them, even when the candidate qualifications were held steady.
Why this matters: If a hiring screen rewards "sounds like this AI wrote it" instead of "best fit for the job," good candidates can get missed for reasons that have nothing to do with ability. That is a hiring quality problem, not just an AI ethics problem.
3. The Workday hiring lawsuit is still shaping the rules
Mobley v. Workday is one of the most closely watched cases about AI in hiring. The plaintiff alleges Workday's screening tools discriminated against applicants based on age, race, and disability. Workday denies the allegations and says its tools do not make hiring decisions and are not trained to use or identify protected characteristics.
Why this matters: The important part for employers is simple: outsourcing the software does not outsource the responsibility. If your company uses a screening vendor, someone still needs to understand what the tool does, what data it uses, and how decisions are checked.
───
🛠️ THIS WEEK'S TRICK
Three Questions to Ask Before You Trust an AI Screener
Most people either adopt AI screening tools without scrutiny or reject them without understanding them. This prompt helps you do neither.
Step 1: Name the tool
Pick the specific AI screening or resume-filtering tool your company uses or is considering. This only works if you're specific.
Step 2: Run this prompt
I'm evaluating whether to use [tool name] to screen candidates for [job title] roles.
Help me think through:
1. What signals is this tool likely using to rank candidates? Consider skills keywords, experience level, job title matches, career path, education, writing style, or anything else that may matter.
2. What types of strong candidates might this tool underrank or miss?
3. What would a simple spot-check look like, comparing the tool's top 10 candidates against my own top 10 from the same applicant pool?
Give me practical answers, not general AI ethics commentary.
Step 3: Do the spot-check
Take the next 20 resumes that come in and rank your top 5 yourself before looking at what the tool produces. Then compare. If your lists are very different, find out why before you scale.
What this works best for:
Managers evaluating a new AI hiring tool for the first time
HR teams that inherited a screening tool and never questioned it
Anyone whose company uses AI screening but nobody clearly owns the decision
Limitation: this won't tell you everything about how the tool works. But it will tell you whether it's producing results you'd actually stand behind.
───
🔧 TOOL OF THE WEEK
Findem — https://www.findem.ai/
Findem is a talent intelligence platform for sourcing and workforce planning. Instead of only searching resumes for keywords, it describes talent through signals like career history, relationships, fit, and role-specific experience.
In the context of this issue, it is a useful example of where AI hiring tools are headed: away from simple keyword matching and toward richer candidate signals. That can be useful, but it also makes the setup questions more important.
Price: No public pricing listed. You have to request a demo.
Best for: Recruiting teams that hire often enough to justify a structured sourcing platform.
Worth trying? Yes, if your team is already doing high-volume sourcing and keyword searches keep surfacing the wrong people. Not worth the sales cycle if you only hire for a few roles a year.
───
💡 PROMPT OF THE WEEK
Use this when you want to write a job description that gives both humans and screening tools a clearer target.
I'm writing a job description for a [job title] role.
The key things this person actually needs to do are:
[list 3-5 real responsibilities]
The strongest signals of success are:
[list 2-4 things, such as outcomes they've driven, work style, domain experience, customer type, or operating environment]
Help me write a clear job description that:
1. Prioritizes the real work over generic requirements
2. Avoids unnecessary filters that could screen out strong candidates
3. Makes the must-have signals easy for a recruiter, hiring manager, or screening tool to identify
Tip: the more specific you are about what the person will actually do, the more useful the output. Generic inputs produce generic job descriptions.
───
👋 THAT'S A WRAP
AI screening tools aren't going away, and that's fine. Speed is genuinely useful when you have 200 applications and three days to fill a role. The problem is using a screener without knowing what it rewards, what it misses, and who is checking the output.
Ask three questions before you hand your candidate pool to a tool: what is it filtering on, what might it miss, and does my own ranking match its ranking? If you don't know the answers, that's where to start.
If you manage a team and someone above you just announced you're "using AI for recruiting now," forward this to them. Not to start a fight, just to make sure someone's asking the right questions.
See you next Friday.