Legal
AI Transparency Statement
Last updated: March 2026
Talent Atrium uses AI to assist with candidate evaluation. This statement explains exactly what AI does in our platform, what data it uses, what it does not do, and where human oversight applies at every stage.
Core principle: AI assists, humans decide
AI is used in this platform to process information faster and more consistently than manual review. It does not make employment decisions. Every shortlisting decision, hiring decision, and rejection decision is made by a human recruiter. The AI provides structured signal; the recruiter provides judgement. This is a non-negotiable design principle of the platform, not a configuration option.
Where AI is used in the platform
CV parsing
When a candidate uploads a CV, the AI reads the document and extracts structured fields including employment history, job titles, dates, skills, and education. This structured output forms the candidate's profile and is used in all subsequent evaluations.
Input data used
CV document content, structured into employment history, education, skills, and dates.
Human oversight
Candidates can review and correct their parsed profile at any time from the candidate portal.
Candidate evaluation
When a candidate applies to a vacancy, the AI evaluates the candidate's structured profile against the vacancy's structured requirements. Each application is assessed across multiple dimensions including experience level, qualifications, skills, and behavioural indicators from the behavioural assessment.
Input data used
Structured profile data, behavioural assessment results, vacancy requirements. Evaluation does not use candidate name or location as scoring inputs.
Human oversight
Scores and reports are presented to the recruiter as a decision-support tool. The recruiter makes all shortlisting and hiring decisions. No candidate is rejected by automated decision alone.
Compatibility report generation
The AI produces a written compatibility report for each candidate that explains, in plain language, how the candidate's profile aligns or does not align with the vacancy requirements. The report references specific dimensions and explains the reasoning behind the score.
Input data used
Structured evaluation outputs from the candidate and vacancy assessment.
Human oversight
Reports are provided to recruiters for review and to candidates upon vacancy closure. They are not binding assessments.
Candidate feedback generation
When a vacancy is closed, the AI generates personalised written feedback for each rejected candidate. This feedback references the candidate's compatibility score, describes the profile of the shortlisted candidates in aggregate, and provides structured guidance on areas for improvement.
Input data used
Evaluation scores, dimension breakdowns, and aggregated shortlist profile data. No personally identifying information from other candidates is included.
Human oversight
Feedback is generated from structured evaluation data and is reviewed for accuracy before being sent. Candidates may contact us to query specific feedback.
Behavioural profile interpretation
Candidates complete a structured behavioural assessment. The AI interprets the responses and produces a plain-language behavioural profile that describes the candidate's workplace tendencies, communication style, and working preferences.
Input data used
Candidate responses to the structured behavioural assessment questions.
Human oversight
Behavioural profile reports are informational only. They are not diagnostic instruments and should not be treated as clinical or psychological assessments.
What AI does not do
- AI does not make hiring decisions of any kind
- AI does not use candidate name or location as scoring inputs in the evaluation model
- AI-generated scores are not final outcomes; they are inputs to human review
- AI does not profile candidates for purposes beyond the stated vacancy evaluation
- The AI processing provider does not retain or train on candidate data submitted through this platform
- AI is not used in account access decisions, fraud detection, or any purpose beyond evaluation and reporting
- No candidate is rejected solely on the basis of an automated score without a human recruiter reviewing and making the decision
Developer statement (Colorado SB24-205)
Talent Atrium has taken reasonable care in the design of this system to protect candidates from algorithmic discrimination. This includes:
- Designing evaluation criteria around objective role requirements rather than demographic characteristics
- Excluding name and location from AI scoring inputs to reduce proxy discrimination risk
- Requiring human recruiter review and decision for all shortlisting and rejection outcomes
- Providing candidates with the right to explanation, correction, and contest of evaluation results
- Maintaining audit records of system operation as required by applicable law
Deployers (recruiters) using this platform are responsible for ensuring their use of evaluation outputs complies with applicable employment and anti-discrimination law in their jurisdiction.
Questions, corrections, and contestation
If you have questions about how AI has been used in your specific evaluation, if you believe an AI-generated report contains an error, or if you wish to contest the outcome of an evaluation, please contact us. We will:
- Acknowledge your request within two business days
- Review the evaluation output in question against the structured data used to produce it
- Provide a written response explaining the basis for the evaluation result
- Notify the relevant recruiter if an error is identified that materially affected a hiring outcome