AI Bias and Automated Employment Decision Tools Disclosure

1. About This Disclosure

Rebecca is an artificial intelligence-powered interviewing and candidate-screening platform operated by Pete & Gabi, a division of Peterson Technology Partners.

This disclosure explains:

• How Rebecca may be used during a recruitment or employment screening process.
• What candidate information Rebecca may process.
• How Rebecca analyzes candidate skills and interview responses.
• The role of employers, recruiters, and human decision-makers.
• How candidates may request an alternative screening method, accommodation, correction, or human review.
• How bias audits and fairness evaluations are conducted and reported.

This disclosure is intended for clients, candidates, applicants, employees, recruiters, employers, staffing agencies, and other individuals who may interact with Rebecca.

The organization that invited you to an interview is referred to in this disclosure as the “Hiring Organization.”

2. What Is Rebecca?

Rebecca is an AI-powered interviewer and recruitment-support platform.

Depending on the Hiring Organization’s configuration, Rebecca may:

• Ask predefined interview questions.
• Generate follow-up questions based on a candidate’s responses.
• Conduct voice or video interviews.
• Record interviews after providing the applicable disclosure and obtaining required consent.
• Convert spoken answers into written transcripts.
• Review resumes, job descriptions, interview responses, and work-related qualifications.
• Compare responses against job-related criteria selected by the Hiring Organization.
• Produce interview summaries.
• Identify topics that were or were not addressed.
• Generate candidate scores, ratings, observations, or recommendations.
• Help recruiters organize and review candidate information.
• Assist in determining which candidates may receive additional human review.

Rebecca does not guarantee that its transcripts, summaries, scores, or recommendations will be complete or correct.

3. Is Rebecca an Automated Employment Decision Tool?

Rebecca may qualify as an automated employment decision tool, high-risk AI system, or automated decision system under certain laws when its output substantially assists or replaces discretionary decision-making in recruitment, hiring, promotion, or another employment decision.

Whether Rebecca is legally classified in this way depends on factors that may include:

• The jurisdiction in which the Hiring Organization operates.
• The candidate’s location.
• How the Hiring Organization configures Rebecca.
• The type of information Rebecca evaluates.
• Whether a score, ranking, classification, or recommendation is generated.
• How much weight the Hiring Organization gives Rebecca’s output.
• Whether Rebecca’s output is used to advance, reject, rank, or otherwise evaluate candidates.

The Hiring Organization is responsible for determining which automated-employment laws apply to its use of Rebecca.

4. How Rebecca Screens Candidate Skills

Rebecca may evaluate information provided by the candidate and the Hiring Organization.

This information may include:

• Resume or curriculum vitae content.
• Employment history and educational background.
• Certifications, licenses, and credentials.
• Professional experience and claimed technical proficiencies.
• Candidate answers to interview questions.
• Work samples or portfolio information.
• Availability and work preferences.
• Responses to technical, behavioral, situational, or communication questions.
• Technical Coding Submissions: When a coding assessment is enabled, live source-code submissions, compiler execution outputs, syntax, algorithmic efficiency, and test-case completions.
• Information contained in the relevant job description.
• Evaluation criteria selected by the Hiring Organization.

Rebecca may compare candidate information with job-related criteria such as:

• Required technical and role-specific knowledge.
• Problem-solving approach and analytical reasoning.
• Technical correctness and code functionality in live coding exercises.
• Experience using particular tools, frameworks, or technologies.
• Familiarity with particular processes or methodologies.
• Completeness and structure of an answer.
• Relevance of the response to the question asked.
• Ability to explain concepts clearly.
• Criteria identified by the Hiring Organization as necessary or preferred for the position.

Rebecca produces a structured assessment showing how a candidate's verbal and technical responses relate to those criteria.

5. How Audio Responses Are Processed

When a voice, video, or technical assessment interview is enabled, Rebecca processes the candidate’s spoken answers and interactive coding submissions through automated evaluation pipelines.

5.1 Audio and Speech Processing

Processing spoken candidate responses may involve:

• Capturing the spoken audio response through the candidate's browser or device.
• Converting spoken language into written text transcripts using automated speech recognition.
• Dividing the response into distinct topics, conceptual components, and answer structures.
• Identifying job-related concepts, technical terminology, and contextual keywords mentioned in the response.
• Comparing the transcribed response against evaluation guides, benchmark answers, and job descriptions configured by the Hiring Organization.
• Assessing whether the response directly addresses the question asked.
• Generating an automated written summary of the candidate's answer.
• Producing a preliminary score, rating, or structured observation based on configured job-related criteria.
• Identifying specific topics, ambiguities, or areas recommended for human recruiter follow-up.

5.2 Technical and Coding Assessment Processing

When a live coding assessment is included, processing technical submissions may involve:

• Capturing real-time source code input, edits, and interactive problem-solving steps within the coding environment.
• Executing submitted source code inside an isolated, secure compiler sandbox.
• Running candidate code against predefined functional test cases, edge cases, and expected output parameters.
• Evaluating syntax accuracy, algorithmic structure, logic flow, and code completeness.
• Recording runtime efficiency, execution errors, compilation warnings, and test-case pass rates.
• Comparing the implementation against expected technical benchmarks and role-specific competencies configured by the Hiring Organization.
• Generating technical observations regarding problem-solving approach, coding proficiency, and debugging methodology.
• Producing structured scoring metrics and code summaries for recruiter review.

5.3 Factors Affecting Processing Quality

The accuracy and quality of automated audio and technical processing may be affected by factors outside the Company’s control, including:

• Internet connection stability, bandwidth constraints, or packet loss.
• Microphone quality, camera performance, or hardware configuration.
• Background noise, acoustic echo, or room interruptions.
• Accents, regional dialects, or non-native speech patterns.
• Speech differences, impairments, or voice modulation characteristics.
• Use of assistive communication technologies or screen readers.
• Language proficiency and communication styles.
• Browser compatibility, local system caching, or network firewall restrictions.
• Compiler execution latency, timeout events, or syntax-parsing limitations.
• Inherent limitations or occasional errors in speech-to-text recognition models.

5.4 Review of Processing and Transcription Errors

A transcription anomaly, acoustic misinterpretation, or compilation issue may result in an incomplete or inaccurate summary, score, or recommendation.

Automated outputs are intended solely to support human decision-making and must not be treated as definitive or infallible determinations of a candidate’s qualifications. Candidates who believe a technical failure, transcription error, or accessibility barrier materially affected their interview or assessment results may request that the Hiring Organization conduct a qualified human review or provide an alternative evaluation method.

6. What Rebecca Is Not Intended to Evaluate

Unless expressly disclosed, legally permitted, and supported by appropriate safeguards, Rebecca is not intended to evaluate candidates based on:

• Race.
• Color.
• Ethnicity.
• National origin.
• Religion.
• Sex.
• Pregnancy.
• Sexual orientation.
• Gender identity or expression.
• Age.
• Disability.
• Medical condition.
• Genetic information.
• Citizenship status.
• Military or veteran status.
• Marital or family status.
• Political beliefs.
• Any other characteristic protected by applicable law.

Rebecca is not intended to determine a candidate’s:

• Emotional state.
• Honesty or truthfulness.
• Mental-health condition.
• Medical diagnosis.
• Personality type.
• Future pregnancy.
• Criminality.
• Cultural suitability.
• Facial attractiveness.
• Likelihood of engaging in misconduct.
• Protected characteristics inferred from a voice, name, appearance, address, or background.

Unless specifically disclosed in a separate legally compliant notice, Rebecca does not use facial recognition or voiceprint identification to identify candidates.

7. Communication-Related Assessments

The Hiring Organization may configure Rebecca to assess job-related communication criteria.

Examples may include:

• Whether the candidate answered the question.
• Whether the response was organized and understandable.
• Whether the candidate explained relevant experience.
• Whether the candidate used appropriate technical terminology.
• Whether the candidate communicated information needed for the position.

Communication assessments should not be used to penalize a candidate solely because of:

• Accent.
• Dialect.
• Speech disability.
• Use of assistive technology.
• Non-native language status.
• Cultural communication style.
• Temporary technical problems.
• Background noise outside the candidate’s reasonable control.

Where communication is evaluated, the criterion should be related to the legitimate requirements of the position.

8. How Scores and Recommendations Are Generated

Rebecca may generate a numerical score, category, ranking, observation, summary, or recommendation.

The output may be based on:

• Criteria selected by the Hiring Organization.
• The candidate’s transcribed answers.
• Resume information.
• The job description.
• Required or preferred skills.
• Ideal-answer or evaluation guides.
• Completion of required interview topics.
• Role-specific competencies.
• Hiring Organization-configured weighting.

A score does not represent an objective or complete measurement of the candidate’s intelligence, personality, value, potential, or future job performance.

A lower score may result from incomplete information, transcription errors, technical problems, evaluation settings, or differences between the candidate’s answer and the configured evaluation guide.

Rebecca’s outputs should be reviewed by an appropriately authorized person before a consequential employment decision is made.

9. Role of the Hiring Organization

The Hiring Organization controls the recruitment process.

The Hiring Organization generally determines:

• Which candidates are invited.
• Which job is being evaluated.
• Which questions can be asked.
• Which skills and qualifications are considered.
• How much weight is given to Rebecca’s output.
• Which recruiters or hiring managers review the results.
• Whether a candidate advances.
• Whether a candidate is offered employment.
• Whether an alternative screening method is available.
• How accommodations and review requests are handled.

Questions about a specific job, decision, score, or rejection should generally be directed to the Hiring Organization.

10. Human Decision-Making

Rebecca is designed to support, not replace, responsible human decision-making.

Hiring Organizations should not treat Rebecca’s output as guaranteed fact.

A human reviewer should consider relevant information such as:

• The candidate’s complete response.
• Resume and employment history.
• Relevant work samples.
• Qualifications required for the position.
• Possible transcription errors.
• Technical issues during the interview.
• Requested accommodations.
• Context that may not have been available to Rebecca.
• Whether the evaluation criteria are job-related.
• Other information obtained through the recruitment process.

The extent of human involvement may vary by Hiring Organization. Candidates may contact the Hiring Organization to ask how Rebecca’s output is used in a particular employment process.

11. Requesting an Alternative Screening Method

A candidate may request an alternative screening or interview process.

An alternative may include:

• An interview with a human recruiter.
• A written-response assessment.
• A telephone interview without video.
• Additional time to complete the assessment.
• Another accessible format.
• Manual review of the candidate’s qualifications.
• A different method offered by the Hiring Organization.

To request an alternative method, contact the Hiring Organization.

Candidates should submit the request as early as reasonably possible and identify the interview or position involved.

A candidate is not required to disclose detailed medical information when initially asking how to request an accommodation or alternative process.

The Hiring Organization is responsible for determining what alternative process can be provided.

12. Requesting an Accommodation

Candidates who have a disability, accessibility need, communication difference, or difficulty using Rebecca may request a reasonable accommodation.

A request may relate to:

• Hearing.
• Vision.
• Speech.
• Mobility.
• Cognitive accessibility.
• Neurodivergence.
• Language processing.
• Use of assistive technology.
• Interview timing.
• Video or microphone requirements.
• Another limitation affecting access to the interview.

Candidates may request an accommodation by contacting the Hiring Organization using the information provided in the interview invitation.

Candidates may also contact the Hiring Organization for technical accessibility questions.

Rebecca may provide technical assistance, but the Hiring Organization is responsible for evaluating and responding to employment-related accommodation requests.

13. Requesting Human Review

A candidate may request human review when the candidate believes that an AI-generated transcript, score, summary, ranking, recommendation, or other output may be materially inaccurate or may have affected an employment decision.

A human-review request may be appropriate when:

• The transcript does not accurately reflect the candidate’s answer.
• The interview was interrupted by a technical problem.
• Audio quality affected speech recognition.
• A disability or assistive technology may have affected the assessment.
• The candidate believes incorrect personal information was used.
• The score appears to be based on an answer to a different question.
• Relevant qualifications were omitted or misunderstood.
• The candidate believes the output was used unfairly.

To request human review, contact the Hiring Organization.

The request should include:

• The candidate’s name.
• The position.
• The approximate interview date.
• The part of the assessment believed to be inaccurate.
• Any relevant explanation or correction.

Candidates should not send passwords, financial information, medical records, or unrelated sensitive information with the request.

14. Requesting Correction of Personal Data

A candidate may request correction of inaccurate personal information used in the recruitment process.

Examples include:

• Incorrect name.
• Incorrect telephone number.
• Incorrect employment history.
• Incorrect education information.
• Incorrect resume version.
• Incorrect interview transcript.
• Incorrect candidate identifier.
• Information belonging to another candidate.

Requests should be directed to the Hiring Organization.

The candidate may be asked to verify their identity before information is corrected or disclosed.

15. Challenging an Adverse Decision

When required by applicable law or offered by the Hiring Organization, a candidate may challenge an adverse decision that was based in whole or in substantial part on Rebecca’s output.

A challenge may include:

• Requesting an explanation of the relevant factors.
• Correcting inaccurate information.
• Providing additional qualifications.
• Requesting reconsideration.
• Requesting human review.
• Requesting another screening method.

The Hiring Organization is responsible for reviewing and responding to employment-decision challenges.

16. Bias and Fairness Risks

AI-supported employment tools may produce unequal or inaccurate outcomes.

Potential sources of bias or unfairness may include:

• Unrepresentative evaluation data.
• Biased job descriptions.
• Biased interview questions.
• Customer-selected criteria that are not job-related.
• Differences in language, accent, or communication style.
• Speech-recognition errors.
• Disability-accessibility barriers.
• Incorrect or incomplete resumes.
• Small sample sizes.
• Differences in how groups are represented in evaluation data.
• Use of proxy variables associated with protected characteristics.
• Human overreliance on automated scores.
• Differences in campaign configuration between candidates.
• Technical differences in devices, microphones, or internet access.

No testing process can guarantee that an AI system will be free from all bias or error.

Fairness requires both technical evaluation and responsible use by the Hiring Organization.

17. Bias Audits

Where required by law or contract, Rebecca or the Hiring Organization may arrange for an independent bias audit of covered automated employment functionality.

A bias audit may evaluate:

• Selection rates.
• Scoring rates.
• Impact ratios.
• Differences across legally specified demographic categories.
• Whether the audited system meets the applicable legal definition of an AEDT.
• The data used in the audit.
• The number of candidates included.
• Excluded or unknown demographic categories.
• The date of the audit.
• The identity and independence of the auditor.
• Limitations of the methodology.
• The specific system version and features tested.

The scope and results of an audit may not apply to every employer, campaign, configuration, job, location, or version of Rebecca.

18. Fairness Evaluations

In addition to legally required bias audits, Rebecca may conduct internal or external fairness evaluations.

These evaluations may examine:

• Speech-transcription accuracy.
• Differences in scoring outcomes.
• Consistency across similar answers.
• False positive and false negative rates.
• Performance across accents or dialects.
• Performance for candidates using assistive technologies.
• Performance across languages.
• The effect of missing or incomplete data.
• Sensitivity to prompt and scoring changes.
• Whether protected characteristics or inappropriate proxies influence outputs.
• Human-review override patterns.
• Candidate complaints and correction requests.
• Differences between AI output and qualified human evaluation.

Internal fairness testing is not necessarily equivalent to an independent bias audit required by law.

19. Bias-Audit Methodology and Limitations

An audit or fairness report should identify material limitations, which may include:

• Limited demographic data.
• Small sample sizes.
• Voluntary or missing demographic information.
• Inability to evaluate intersectional groups.
• Differences between historical data and future applicants.
• Differences among job categories.
• Differences among employer configurations.
• Changes to the system after the audit.
• Incomplete outcome information.
• Data-quality issues.
• Legal limitations on collecting protected-characteristic data.
• Dependence on customer-provided information.
• Differences between scoring and final employment decisions.

An audit should not be interpreted as a guarantee that discrimination cannot occur.

20. Employer-Specific Audits

A platform-level audit may not fully evaluate how a specific Hiring Organization uses Rebecca.

Employer-specific risks may be affected by:

• Job descriptions.
• Interview questions.
• Scoring guides.
• Weighting.
• Candidate population.
• Geographic location.
• Job requirements.
• Human-review procedures.
• Selection thresholds.
• Other screening tools.
• Historical employment practices.
• How Rebecca’s output is combined with other information.

Hiring Organizations may need to conduct their own impact assessments, bias audits, validation studies, or legal reviews.

21. Protected Characteristics and Audit Data

Rebecca does not require candidates to provide demographic information unless the collection is:

• Lawful.
• Voluntary where required.
• Properly disclosed.
• Separated from hiring decision-makers where appropriate.
• Used for an authorized compliance, reporting, or fairness purpose.
• Subject to appropriate security and access controls.

Information collected for a bias audit or fairness evaluation should not be used to make an employment decision about the candidate.

22. Data Used by Rebecca

Depending on the Hiring Organization’s settings, Rebecca may process:

• Candidate name.
• Telephone number.
• Email address.
• Resume.
• Employment history.
• Education.
• Certifications.
• Interview audio.
• Interview video.
• Interview transcript.
• Interview summary.
• Candidate responses.
• Job description.
• Hiring criteria.
• Scores and summaries.
• Technical and device information.
• Interview timing and connection information.

Additional information about data collection, retention, sharing, and candidate rights is available in the Rebecca Privacy Policy:

Privacy Policy: https://www.petegabi.com/privacy-policy/

23. Data Retention

Candidate Personal Data, interview recordings, transcripts, coding submissions, summaries, and assessment scorecards are retained during the Hiring Organization’s active subscription.

Applicable Retention Period:

Records are ordinarily deleted from active production systems within ninety (90) days following account closure or agreement termination, unless a custom retention period is configured by the Hiring Organization or earlier deletion is requested.

Candidates may contact the Hiring Organization to request information about:

• The categories of data collected.
• The source of the data.
• The retention period.
• The organization’s deletion process.
• The parties that may receive the information.

Deletion rights may be subject to legal, regulatory, litigation, fraud-prevention, or record-retention obligations.

24. Notice Before Use

Where required, candidates will receive notice before Rebecca is used to assess or evaluate them.

The notice may identify:

• That an automated tool will be used.
• The name of the tool.
• The job qualifications or characteristics being assessed.
• The categories of data collected.
• The source of the data.
• The applicable retention policy.
• How to request an alternative process.
• How to request an accommodation.
• How to ask questions.
• How to access published audit information.

The Hiring Organization is responsible for providing notices within legally required timeframes.

25. No Retaliation

Candidates should not be retaliated against solely for:

• Requesting an accommodation.
• Requesting an alternative interview.
• Asking whether AI is being used.
• Requesting human review.
• Reporting an inaccurate transcript.
• Requesting correction of personal data.
• Asking for bias-audit information.
• Exercising a privacy right.
• Reporting suspected discrimination.
• Participating in an investigation.

Questions or concerns about retaliation should be directed to the Hiring Organization.

26. Reporting Bias or Discrimination Concerns

Candidates may report a concern involving:

• Discriminatory questions.
• Inaccurate transcription.
• Unfair scoring.
• Accessibility barriers.
• Failure to provide an alternative process.
• Improper use of a protected characteristic.
• Failure to provide required notice.
• Use of an unaudited AEDT where an audit is required.
• Failure to provide human review where required.
• Another suspected violation.

Reports may be submitted to:

Hiring Organization.

Include the position, interview date, Hiring Organization, and a description of the concern. Do not include passwords or unrelated sensitive information.

27. Investigation and Remediation

When Rebecca receives a credible report, it may:

• Review relevant logs and system outputs.
• Review the transcript or recording where authorized.
• Evaluate the system configuration.
• Ask the Hiring Organization for information.
• Correct technical errors.
• Refer employment decisions to the Hiring Organization.
• Suspend a feature or campaign where necessary.
• Conduct additional fairness testing.
• Update documentation or controls.
• Preserve information required for an investigation.

Rebecca cannot guarantee that the Hiring Organization will reverse or change an employment decision.

28. Changes to Rebecca

Rebecca’s models, prompts, scoring processes, integrations, and system features may change over time.

A material change may affect whether prior audit or fairness results continue to apply.

Rebecca should maintain records identifying:

• The version evaluated.
• The date the version was placed into use.
• Material changes after the evaluation.
• Whether a new audit or evaluation was conducted.
• The date updated results were published.

30. Important Notice

Rebecca provides technology that assists Hiring Organizations with interviewing and candidate evaluation.

The Hiring Organization remains responsible for:

• Determining whether an AEDT or high-risk AI law applies.
• Providing required candidate notices.
• Arranging any employer-specific bias audit.
• Providing reasonable accommodations.
• Offering an alternative process where required.
• Providing human review.
• Ensuring evaluation criteria are job-related.
• Preventing unlawful discrimination.
• Making the final employment decision.
• Responding to candidate questions about a specific job or decision.

This disclosure does not provide legal advice and does not guarantee compliance with every law or regulation.