Privacy and Security
for Artificial Intelligence.

privacy

Differential privacy budgets in LLM fine-tuning: a practitioner's guide

A reproducible study of ε–δ budgets, gradient clipping, and per-sample noise on three open-weight models, with concrete production trade-offs.

Mar 2026

security

Membership inference attacks on retrieval-augmented generation

We show that RAG pipelines leak the presence of specific documents in the index under realistic prompting conditions, and propose two practical mitigations.

Jan 2026

governance

Toward a baseline for AI incident disclosure: lessons from CVD in software

A proposal for coordinated AI incident disclosure modelled on RFC 9116 and ISO/IEC 29147, mapped onto the NIST AI RMF.

Nov 2025

Methodology

AI Privacy Risk Assessment (APRA) — v1.0

An open methodology for assessing privacy risk across the AI lifecycle. Combines NIST Privacy Framework outcomes with LINDDUN threat modelling.

Methodology

ML Red-Team Playbook — v1.0

Operational guide for adversarial testing of ML systems. Covers test design, evidence collection, and reporting aligned to OWASP LLM Top 10.