Uploaded June 2026 | Updated September 2026, 3 weeks ago
PEPR '26 - Turning Privacy Risk Assessment Into 20 Questions for Developers
Qiyu Li, University of California, San Diego
Privacy design is often treated as inherently open-ended and deeply contextual. Small shifts in context—who requests data, for what purpose, under what stakes, and with what downstream sharing—can flip whether a data practice feels appropriate. While this captures a real phenomenon, it has also left us with privacy frameworks that are too abstract and generic to operationalize for developers without specialized privacy expertise.
In this talk, we explore methods to significantly lower the barrier of privacy risk assessment for developers. We argue that privacy design is more structured than it appears: many recurring privacy-relevant decisions can be captured using a structured, closed-ended representation. By constructing such a representation, we can transform a significant portion of open-ended privacy design into a closed-ended task.
We operationalize this through PrivacyAkinator, an interactive tool that guides developers through key privacy design decisions via LLM-generated multiple-choice questions. By grounding privacy assessment in a structured representation of the privacy design space, PrivacyAkinator transforms privacy design from ad-hoc decision-making into systematic, auditable design support.
Authors: Qiyu Li, Haojian Jin
View the full PEPR '26 program at usenix.org/conference/pepr26/program
PEPR '26 - Turning Privacy Risk Assessment Into 20 Questions for Developers
Qiyu Li, University of California, San Diego
Privacy design is often treated as inherently open-ended and deeply contextual. Small shifts in context—who requests data, for what purpose, under what stakes, and with what downstream sharing—can flip whether a data practice feels appropriate. While this captures a real phenomenon, it has also left us with privacy frameworks that are too abstract and generic to operationalize for developers without specialized privacy expertise.
In this talk, we explore methods to significantly lower the barrier of privacy risk assessment for developers. We argue that privacy design is more structured than it appears: many recurring privacy-relevant decisions can be captured using a structured, closed-ended representation. By constructing such a representation, we can transform a significant portion of open-ended privacy design into a closed-ended task.
We operationalize this through PrivacyAkinator, an interactive tool that guides developers through key privacy design decisions via LLM-generated multiple-choice questions. By grounding privacy assessment in a structured representation of the privacy design space, PrivacyAkinator transforms privacy design from ad-hoc decision-making into systematic, auditable design support.
Authors: Qiyu Li, Haojian Jin
View the full PEPR '26 program at usenix.org/conference/pepr26/program
