Pwning ML for Fun and Profit

Presented at Kiwicon X: The Truth is In Here (2016), Nov. 18, 2016, 2:15 p.m. (45 minutes).

Everyone is talking ML this and AI that as if they expect some kind of Utopian beast to be waiting just behind the next door and whisk us all away to a technological-paradise. It would seem dire warnings of every Sci-Fi book and movie ever haven't been enough to dissuade people from cooking statistics and math into an techno-optimist soup of dubious origin and expecting us to swallow. Obviously security can't just sit here and watch the catastrophes unfold. I aim to lay out some of the most awful yet still amusing examples of how and why we can and will break things. This presentation attempts to offer the audience a refreshingly realistic look at the terrible flaws in ML, the ease of altering outcomes and the dangers ahead..


Presenters:

  • Davi Ottenheimer
    Winner of the KiwiCon7 "best dressed" award. Also president of flyingpenguin, co-author of the book "Securing the Virtual Environment: How to Defend the Enterprise Against Attack," and author of upcoming book about the infrastructure flaws behind learning systems: "The Realities of Securing Big Data".

Links:

Similar Presentations: