Tootfinder

Opt-in global Mastodon full text search. Join the index!

@privacity@social.linux.pizza
2026-02-25 18:34:23

Deanonimizzazione online su larga scala con LLM
Gli LLM possono essere utilizzati per eseguire una deanonimizzazione su larga scala. Con un accesso completo a Internet, il nostro agente può reidentificare gli utenti di Hacker News e i partecipanti ad Anthropic Interviewer con elevata precisione, dati solo profili e conversazioni online pseudonimi, un'operazione che richiederebbe ore a un investigatore umano dedicato. Progettiamo quindi attacchi per l'ambiente chiuso. Dati due d…

Large-scale online deanonymization with LLMs
We show that large language models can be used to perform at-scale deanonymization. With full Internet access, our agent can re-identify Hacker News users and Anthropic Interviewer participants at high precision, given pseudonymous online profiles and conversations alone, matching what would take hours for a dedicated human investigator. We then design attacks for the closed-world setting. Given two databases of pseudonymous individuals, each containing unstructured text written by or about tha…

@chiraag@mastodon.online
2026-02-21 06:34:15
Content warning: LLMs, creativity, social consequences

This will be a (long-ish) thread about LLMs, particularly in response to a few different conversations I've either been a part of or witnessed recently.
So many people seem to have trouble grasping that their usage of LLMs, even within a limited context, inherently normalizes *all* contexts for LLMs, whether they mean to do so or not.
1/?

@tante@tldr.nettime.org
2026-02-22 18:38:39

Many of the people I do have online conversations with speak English as their first language. I kinda get by but you native speakers really underestimate how hard it is to express oneself in a foreign language. Concepts work differently, metaphors don't really translate, references you have used for decades don't make sense.