Weaponizing Data Science for Social Engineering: Automated E2E Spear Phishing on Twitter
summary
An end-to-end system called Snapper clusters Twitter profiles to pick high-value targets, scrapes timelines for topics and active hours, then seeds a recurrent neural network or a Markov chain to generate @-reply tweets ending in unique goo.gl short links. One instance posts ~819 tweets in two hours and records 275 click-throughs versus a human’s 49; pilot tests show about 30% click-through conservatively (up to 66% including bot noise). The tool obeys rate limits and profiles timing to increase success.
AI-written summary · about
chapters (21)
- 0:00Intro
- 1:05A Novel Phishing Campaign Design
- 2:32Fooling Humans for 50 Years
- 6:30ISO: Demo Volunteers
- 7:23Why Twitter?
- 9:47Techniques, Tactics and Procedures
- 12:16Design Flow
- 13:02Triage of High Value Targets on Twitter
- 18:03Choosing a URL shortener
- 19:09Desirable properties of URL shortener
- 20:08Recon and Footprinting for Profiling
- 21:47Leveraging Markov Models
- 23:13Training a Recurrent Neural Network
- 24:59Tradeoffs and Caveats
- 26:56Language and Social Network Agnosticism
- 28:32Wild Testing SNAP R
- 29:13Pilot Experiment
- 30:33Man vs. Machine 2 Hour Bake Off
- 34:48Potential Use Cases
- 36:01Mitigations
- 37:17Black Hat Sound Bytes
related talks
07mSpear Phishing the Masses: When Open Data is Dark
50mCatPhish Automation - The Emerging Use of Artificial Intelligence in Social Engineering
55mSocial Engineering Overview-How To Prevent Your Co. from Getting Hacked
25mTesting Your Organization's Social Media Awareness
27mCatching the Phisherman
32mFriend or Replicant: How Attackers Automate and Disguise Themselves
44mHacking Humans: Social Engineering Techniques
30mSecurity Starts With You: Social Engineering