PEPR '24
as the organisers put it
The 2024 USENIX Conference on Privacy Engineering Practice and Respect (PEPR '24) will take place on June 3–4, 2024. PEPR is focused on designing and building products and systems with privacy and respect for their users and the societies in which they operate. Our goal is to improve the state of the art and practice of building for privacy and respect and to foster a deeply knowledgeable community of both privacy practitioners and researchers who collaborate towards that goal.
who it's for
privacy engineers, data engineers, data scientists, machine learning engineers, product managers, legal and compliance teams, privacy researchers
topics
differential privacy mechanisms; privacy budgets and accounting; noisy SQL aggregates in Presto; synthetic data evaluation metrics; privacy-preserving MPC and TEE workflows; distributed differential privacy for ad attribution; event-level deletion and data deletion at scale; automating privacy reviews with large language models; consent management and global privacy control; anonymization threat modeling and practices; privacy-preserving federated learning; privacy testing with differentially private synthetic data
| title | author | length |
|---|---|---|
| Empirical Privacy Metrics: The Bad, the Ugly… and the Good, Maybe? | Damien Desfontaines | 14m |
Synthetic-data vendors often report similarity-based metrics that compute a distance-to-train vs distance-to-test and flag overfitting, but those numbers are… | ||
| Balancing Privacy: Leveraging Privacy Budgets for Privacy-Enhancing Technologies | Jordan Brandt | 16m |
Argues that cryptographic PETs (MPC, FHE, TEEs) protect input but not output privacy and proposes concrete mitigation: treat mixed deployments as hybrid PETs… | ||
| Utility Analysis for Differentially-Private Pipelines | Vadym Doroshenko | 13m |
An open‑source Python module inside PipelineDP models how contribution bounding and Laplace noise trade off to affect accuracy, estimating error per… | ||
| Presto-Native Noisy Aggregations for Privacy-Preserving Workflows | Kien Nguyen, Chen-Kuei Lee, Jonathan Hehir | 10m |
At Meta, Presto gains built-in DP primitives so SQL returns noisy aggregates instead of raw counts, enabling privacy across billions of users while keeping… | ||
| Designing for User Privacy: Integrating Differential Privacy into Ad Measurement... | Jian Du, Shikun Zhang | 20m |
A practical system integrates distributed differential privacy with MPC/TEE-based matching to protect both intermediate and final ad-attribution outputs… | ||
| Being Intentional: AB Testing and Data Minimization | Matt Gershoff | 19m |
Claims that standard AB-test inference (t‑tests, ANOVA, nested partial F‑tests) can run without per‑user records by storing only counts, sums, sums of squares… | ||
| FTC, Privacy, and You | Jessica Colnago | 12m |
The FTC enforces consumer protection and competition with casework, rulemaking, and technical staff who advise investigations. Examples: BetterHelp shared… | ||
| Don't End Up in the PETs Cemetery | Simon Fondrie-Teitler | 12m |
Privacy enhancing technologies sit on a spectrum from policy‑reliant controls to cryptographic guarantees; multi-party computation is used for analytics by… | ||
| Anonymization Aspects of a Low-latency VoIP Security Analytics System | Jiri Kuthan, Cristian Constantin | 15m |
A low-latency VoIP analytics pipeline encrypts only PII—encrypt PII such as SIP URIs, E.164 numbers, IP addresses—end-to-end from source probe to client GUI… | ||
| How We Can Save Anonymization | Daniel Simmons-Marengo | 15m |
Argues that effective anonymization exists and gives concrete checks: skepticism, rigorous threat models that include grad students and intimate‑partner… | ||
| A New Model for International, Privacy-Preserving Data Science | Curtis Mitchell | 11m |
A phase‑one pilot demonstrates a remote, privacy‑preserving workflow that uses PySyft with nodes hosted at NSOs and a network gateway at the UN PET Lab to run… | ||
| Compute Engine Testing with Synthetic Data Generation | Jiangnan Cheng, Eric Liu | 16m |
Meta builds a Synthetic Query Bank that runs privacy-safe, production-like tests against Presto (classic, Velox, Presto on Spark) to detect correctness… | ||
| Empowered User Control: Learnings from Building "Event Level Deletion" | Lingwei Meng, Yan Li | 15m |
Meta implements event-level deletion for ad interactions by storing each click as a single heterogeneous blob keyed with a unique ID and a lightweight… | ||
| Lineage Quality Measurement | JiSu Kim, Alex Lambert, Francesco Logozzo | 13m |
Meta builds a Lineage Quality Measurement framework that quantifies lineage tools by recall and precision metrics, using sampled ground truth from the… | ||
| Deleting Data at Organizational Scale | Diogo Lucas | 16m |
Stripe shows a working deletion pipeline that removes many billions of documents deleted across hundreds of thousands of datasets and hundreds of petabytes… | ||
| Building Permissions into Data Modeling | Lingtian Cheng | 14m |
A production permission framework at Meta combines a rule engine that evaluates named rules (with optimization like parallel execution and ordering to run the… | ||
| Lessons Learned Deploying the Global Privacy Control to Millions of Customers | Ryan Guest | 11m |
Amazon reports rolling Sec-GPC to millions and finds the Global Privacy Control is just an HTTP header with a 0/1 boolean but behaves messily: clients send… | ||
| Consent Management at Airbnb | Aziel Epilepsia, Fernando Rubio | 13m |
Airbnb builds a configuration-driven Consent Management Platform that models terms (strictly necessary, functional, targeting, performance), stores immutable… | ||
| Approaches and Challenges to Purpose Limitation across Diverse Data Uses | Rituraj Kirti, Diana Marsala | 14m |
Meta describes a working system that uses annotations and lineage to label assets, traverse data flows, and apply policy checks; engineers run code inside… | ||
| Governing Identity, Respectfully | Wendy Seltzer | 15m |
User-centric identity efforts keep failing at scale: dozens of self-sovereign projects failed and personal data stores never arrived. Governance must pair… | ||
| Through the Lens of LLMs: Unveiling Differential Privacy Challenges | Aman Priyanshu, Yash Maurya, Vy Tran | 10m |
LLMs are used to simulate membership inference and reidentification attacks against Google’s Topics API, finding that a hierarchical model identifies noisy… | ||
| Learning and Unlearning Your Data in Federated Settings | Tamara Bonaci | 15m |
Federated unlearning requires methods that approximate full retraining because exact machine unlearning is impractical at scale; the proposal analyzes multiple… | ||
| Navigating the Privacy Landscape of AI on the Devices: Challenges and Best Practices | Tousif Ahmed, Mina Askari, Joe Genereux | 15m |
On-device generative AI introduces concrete risks: models ingest camera, microphone, motion sensors data used for recorder summarization, visual assistants and… | ||
| Panel: Privacy Design Patterns for AI Systems: Threats and Protections | Debra J Farber, Engin Bozdag, Vasudha Hegde, Sri Pravallika | 43m |
Panelists lay out concrete defenses and gaps: build an inventory, tag data at collection, and run anomaly detection and red‑team/pen‑test campaigns to find… | ||
| Automating Technical Privacy Reviews Using LLMs | Stefano Bennati, Engin Bozdag | 16m |
Proof-of-concept assesses 80 HERE and 60 Uber architecture documents (half labelled containing at least one privacy risk) using GPT-4 and locally deployed… | ||
| Designing a Data Subject Access Rights Tool | Arthur Borem, Elleen Pan, Olufunmilola Obielodan, Aurelie Roubinowitz, Luca Dovichi, Sophie Veys, Daniel Serrano, Madison Stamos, Margot Herman, Nathan Reitinger, Michelle L. Mazurek, Blase Ur | 11m |
Study reports focus groups, a quantitative survey and analysis of 800 de-identified files and builds an interactive web tool that parses JSON/CSV/HTML exports… | ||
| Building a Protocol to Improve DSR Flexibility and Integration | Prachi Khandekar, Sam Alexander, Suejung Shin | 11m |
A concrete open source protocol implements the Data Rights Protocol: JSON payloads with three required blocks—data subject identity, controller metadata, and… | ||
| Why "Safety"? How We Use Measurement to Advance Governance of Technology | Irene Knapp | 18m |
Internet Safety Labs presents an independent 501(c)(3) product-safety testing approach that reframes privacy work as safety and builds measurable standards… | ||
| Measuring Privacy Risks in Mobile Apps at Scale | Lisa LeVasseur, Bryce Simpson, Irene Knapp, Saish Mandavkar, George Vo | 17m |
Internet Safety Labs collects 88,000+ app data points, captures network traffic for ~1,300 apps, and scores 1,357 apps using programmatic company/SDK/subdomain… | ||
| "A Big Deal": How Privacy Engineering Can Streamline the M&A Process | Shlomi Pasternak | 19m |
Privacy M&A integrates hands‑on privacy engineering with ML and generative tools to speed pre‑close and post‑close assessments, finding concrete issues like… | ||
| Cache-22: Doing Privacy Engineering with Privacy Standards | Zachary Kilhoffer, Devyn Wilder | 16m |
Fourteen hourlong interviews reveal that privacy engineers split between compliance as baseline (GDPR, COPPA) and occasional use of frameworks (NIST, ISO… | ||
| nothing matches | ||