AI & Automated Decision-Making Notice

1. AI Features We Provide

• AI-assisted question generation and question editing help for employers (plan-gated).
• AI grading of short and long written answers when an assessment enables AI grade (plan-gated).
• On-device proctoring models for face presence, gaze patterns and voice comparison.
• Rule-based integrity event detection and an integrity summary band for reviewers.
• Resume parsing and ATS-style skill match scores for sorting (not LLM hiring verdicts).

2. How AI Is Used

Cloud language-model providers (when configured for the deployment — for example OpenAI, Google Gemini or DeepSeek) receive text prompts such as topics, question drafts, candidate written answers and rubrics. On-device models run in the candidate’s browser for camera and microphone signals and do not upload face templates as cloud biometric search indexes.

3. AI-Assisted Candidate Evaluation

AI grading proposes scores for eligible written answers. Coding questions are primarily graded by a sandbox running tests. Multiple-choice and similar types are auto-marked by fixed keys. Recruiters can override grades.

4. Automated Recommendations and Scoring

The Platform may show scores, integrity bands, resume match percentages and stage suggestions. These are decision-support tools for the employer, not a binding automated refusal or hire by Eroforze Systems Private Limited.

5. Human Review and Employer Decision-Making

Hiring, rejection and interview decisions are made by the employer customer. Employers should review AI outputs and integrity evidence with human oversight, especially where law restricts solely automated decisions.

6. AI Proctoring and Anomaly Detection

Proctoring features may flag no face, multiple faces, looking away patterns, other voices, virtual cameras, automation or debugger attachment. Flags create events and evidence for human review. They are not a conclusive finding of cheating by themselves.

7. Limitations and Accuracy

AI systems can be wrong, biased or incomplete. Browser monitoring cannot see phones outside the camera view or applications when the screen is not shared. Do not treat AI grades or integrity bands as ground truth.

8. Human Intervention and Review

Candidates who believe an AI grade or integrity flag is wrong should contact the hiring organisation to request human review. Workspace reviewers can regrade answers and interpret integrity reports.

9. Candidate Rights

Where GDPR or similar laws apply to automated decision-making, candidates may have rights to obtain human intervention, express a view and contest a decision. For hiring outcomes, exercise those rights with the employer (controller). For platform account data, contact privacy@edgeryt.com.

10. Data Used by AI Systems

• Cloud LLMs: text of questions, answers and rubrics as needed for the feature.
• On-device proctoring: camera/microphone streams processed locally for checks described before start.
• We do not send raw assessment video/audio recordings to LLM providers for grading in the current product.

11. Third-Party AI Providers

Optional providers are listed in our Subprocessor List. Only providers configured with API keys on the deployment process data.

12. Contact and Complaints

Platform questions: privacy@edgeryt.com. Hiring decision complaints: contact the employer shown on your application or invite. See also the Privacy Policy.