Creepy stuff with anonymous data
summary
Claims that anonymized posts are reidentifiable using mass-scale text analysis and simple metadata. The talk shows concrete signals — timezone and posting cadence, emoji order, consistent misspellings and phrasing — that form a unique fingerprint. Data sources include commercial brokers or botnet harvesting of public Wi‑Fi and API traffic. Scale is inexpensive: trillions of points for under $100k or a few AWS instances. Results produce probabilistic gender/timezone/person matches good enough for targeted blackmail, and regulations like GDPR do little against abuse.
AI-written summary · about
related talks
40mUsing Stylometry and Machine Learning...
32mThe Dangerous Data Anonymization - Keynote
39mAccidental Abyss: Data Leakage on The Internet
07mSpear Phishing the Masses: When Open Data is Dark
29mPersonal Data Pseudonymization Versus Anonymization In The Age Of AI & Big Data
26mCan You Keep A Secret? Or Privacy In The Digital Age
01h 17mTo be or not to be (anonymous)? Anonymyity in the age of big and open data.
33mPractical Data Privacy