Blog
Notes from the DeepSuck open-source anti-scam project: progress, technical write-ups and scam-recognition guides.
-
Why we use AI to keep scammers busy
DeepSuck has a simple goal — let scammers spend their time on someone who will never fall for it, and keep the clues that help warn others. Here is why we do it, and what we will never do.
-
Inside Auntie Wang's voice pipeline
A scammer says one sentence in the browser or over the phone. Auntie Wang has to answer within a second or two while quietly noting and masking any card number or URL. This post follows one turn through speech recognition, the language model, speech synthesis, fillers, barge-in and transport, plus the rules for extracting and masking intelligence.
-
Warning signs of seven common scam scripts
Scam scripts vary endlessly, but they follow the same arc — a pretext, then trust-building, pressure, and finally a demand for money or information. This guide lists the warning signs and responses for seven common types of fraud in China. It explains how to recognise them and deliberately does not reproduce any lines that could be reused.