Defensive Publication v2.0 — Multi-Layer PII Defence System for AI-Assisted Recruitment Assessment

This is the authors' abstract. We don't add key points for this paper.

Defensive publication establishing prior art for a six-layer personally identifiable information (PII) defence pipeline designed for AI-assisted recruitment assessment. The system combines adversarial normalisation, regex pattern detection, heuristic named entity recognition, confidence-weighted scoring, semantic redaction with industry-aware classification, and bidirectional AI gateway validation. This architecture operates before and after AI model inference to prevent PII exposure during candidate assessment. Published by Assayd Ltd to establish prior art and prevent third-party patenting of these techniques.