AI Hiring Compliance
How Screened supports AI hiring compliance
Set screening criteria, control automations, and review the reasons behind recommendations. Access screening records and testing documentation for leadership and legal review.
Screening starts with your criteria
Your team sets the must-have and nice-to-have requirements for each job. Screened uses these criteria to assess resumes, rather than learning from past hiring decisions.
Control screening criteria and automated actions
You choose the score that counts as a strong fit, and you can set a different one for each job.
Choose which actions to automate and review candidates whenever you need to. You can move someone forward even if the screening recommendation is Not a Fit. Read about human review and automated screening decisions.
Understand AI screening recommendations
Each requirement comes back marked found, unclear, or not found, with a written explanation based on the resume.
See which requirements supported the recommendation and where information was missing or unclear.
Review interview answers and assessments
Your team sets the interview questions and what to look for in the answers. Follow-up questions can vary with the candidate’s responses.
Read the transcript, review the assessment or listen to the recording to understand an interview recommendation. Your team can return to the candidate’s answers when discussing the next step.
Described a refund dispute they escalated, and what changed afterwards.
Screening records and audit trails
Review screening results, written reasons and candidate activity in one place. Export the latest decisions to support internal reviews and show who made each decision and when.
For a wider review, the account audit log records changes to your team, integrations and exports. The screening records guide explains which records to use and what each export includes.
| Candidate | Moved to | By |
|---|---|---|
| Shortlisted | Lena | |
| Interview booked | Sam | |
| Archived | Automation |
What the AI is told to ignore
Screened’s instructions tell the AI to ignore personal details such as name, gender, and age when assessing fit for the role.
Our monthly checks test whether changing those details affects recommendations. The AI bias reports explain the comparisons, results, and limitations.
Bias and consistency, tested monthly
We use fictional resumes and interview transcripts to test whether a name or personal detail changes a recommendation. Published reports describe the results and limits of those internal tests.
Understand your responsibilities as an employer
AI hiring requirements depend on where and how you recruit. Your organisation remains responsible for candidate notices, human review, recordkeeping and any independent audit required for your use of AI.
Use the AI bias reports, system documentation and candidate notice alongside your own policies. The employer duties guide lists what applies where, and when. You can also read our guides to NYC Local Law 144 and the EU AI Act and AI screening.
AI hiring documentation for legal review
Share these with leadership, legal and procurement to review how screening works, what we test and how candidate data is handled.
Common questions
Screened assesses candidates against your job requirements and is instructed to ignore personal details such as name, gender and age. Monthly tests check whether changing those details affects recommendations. Your team can review the results and the reasons behind them; testing helps identify risks but cannot rule out every source of bias.
Yes. Open the candidate’s profile, review the reasons and move them forward if you disagree with the recommendation. Your team chooses what to automate and when to step in.
Yes, when your team enables the relevant rules. These include invitations and shortlisting. You can also set application questions to mark candidates Not a fit when their answers do not meet your criteria.
Screened-hosted application pages explain how AI is used, and the interview invitation introduces Ryan as an AI recruiter. Ryan also introduces itself during the call. For candidates applying through your ATS, include an AI notice in your own application process.
Yes. Ryan greets the candidate by name and introduces itself as an AI recruiter. The assessment focuses on their answers and the job requirements.
Screened compares candidates with the requirements your team sets. Its safeguards and regular testing support job-relevant assessments, and your team can review the reasons behind each recommendation.
Our AI providers do not train on customer data. Public testing uses fictional resumes and interviews. Account-level testing uses your own applications when the Audit add-on is enabled. Read about AI providers and candidate data processing.
Screened runs internal tests to check screening behaviour. These tests are not an independent audit or a compliance certification. Reports can support an external review, but do not replace an independent bias audit where one is required. Read about internal testing and independent reviews.
View our published AI bias reports. Each report explains that month’s tests, results, limitations, and findings. Loading published results.
Published monthly reports record material changes to screening models or instructions. Review the changes alongside the test results and limitations to understand their impact.
Yes, as an add-on. The same tests run inside your account on your own applications, with results in your dashboard and a report you can share. Nothing is written to a candidate's record.
Export candidate records for the latest screening decisions, including who made them and when. Use the account audit log to review team changes, integration connections and exports. For a complete review, include candidate activity and any decisions recorded in your ATS.
You choose how long to keep candidate records and interview recordings. Candidate records are kept for 48 months by default; recordings have a separate setting. See candidate data retention in Screened for settings and deletion options.
These results show how the system behaved on fictional test profiles under controlled test conditions at the time of testing. They come from Screened’s internal tests and do not establish legal compliance. They are not an independent audit, a certification or a New York City Local Law 144 bias audit.