Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

PersonaFingerprint: Measuring Persona Inference on Modern Websites with LLM-Driven Browsing

About

Website Fingerprinting (WFP) has traditionally focused on inferring which website a user visits from encrypted traffic metadata such as packet sizes and timing. In this paper, we identify and quantify a new privacy risk in modern web settings: an adversary can infer a user's persona using only packet-length and inter-arrival-time sequences. To study this risk at scale, we build an LLM-driven multi-agent browsing framework that enforces controllable persona constraints while a computer-use agent interacts with real websites and collects corresponding encrypted traffic traces. We formalize persona fingerprinting under both closed-set and open-world settings and further evaluate whether persona information is already embedded in representations learned by existing WFP models and can be amplified at low cost. Across 10 modern websites and 15 personas (plus an open-world class), persona inference achieves about 84% accuracy on mixed-site traffic; moreover, a lightweight multi-task objective can boost persona accuracy to around 80% while retaining strong site classification performance (about 93% baseline). Our results show that, on modern websites, encrypted traffic metadata can leak not only which site a user visits, but also how they browse and who is browsing.

Chuxu Song, Hao Wang, Richard Martin• 2026

Related benchmarks

TaskDatasetResultRank
Persona FingerprintingAMAZON
OW Precision70.8
1
Persona FingerprintingYouTube
OW Precision66.1
1
Persona FingerprintingREDDIT
OW Precision64.9
1
Persona Fingerprintingcnn
OW Precision62.3
1
Persona FingerprintingYelp
OW Precision71.6
1
Persona FingerprintingBilibili
OW Precision59.7
1
Persona FingerprintingeBay
OW Precision68.9
1
Persona FingerprintingYahoo
OW Precision63.8
1
Persona FingerprintingZhihu
OW Precision61.5
1
Persona FingerprintingLinkedIn
OW Precision67.2
1
Showing 10 of 10 rows

Other info

Follow for update