Hacking AI Models

Presented at Disobey 2024, Feb. 17, 2024, 2 p.m. (180 minutes).

This workshop is about hacking the AI. We will first go through the fundamentals of responsible AI practices, including security testing basics & principles. We are going to take the hacker's approach to AI use cases and see if we can make them do something they aren't supposed to. Participants will have the opportunity to test out a few models from beginner to advanced level.

Presenters:

• Riku Juurikko
  Riku works as Head of Preparedness & Cyber Resilience at Elisa. He's active member of cyber community in Finland being involved with HelSec, Next Gen Hack FI, Women4Cyber Finland.
• Ilja Ikonen
  Dungeon Master who works and plays golf.
• Svitlana Chaplinska
  Svitlana Chaplinska is an information security consultant at Fraktal and HelSec volunteer, who is committed to making security accessible and understandable for everyone. With a keen interest in the evolving role of AI in cybersecurity, she leverages her creative perspective and enthusiasm to drive positive change in the industry.
• Ville Päivinen
  Cybersecurity geek, student, wannabe community hero. Day job Cyber @MSFT

Links:

• https://disobey.fi/2024/profile/disobey2024-198-hacking-ai-models

Similar Presentations:

• Black Hat Europe 2017 / The Machine Fights Back — Leveraging AI for Self-Learning Cyber Security
• Black Hat USA 2020 Virtual / Superman Powered by Kryptonite: Turn the Adversarial Attack into Your Defense Weapon
• CactusCon 11 (2023) / Will AI Take My InfoSec Job? Exploring interesting AI use cases