Colloquium on Cybersecurity/Cyberlaw—Sponsored by the Orrick Fund
In Not With A Bug, But With A Sticker, Ram Shankar Siva Kumar and Hyrum Anderson propose that the most significant risk to currently deployed artificial intelligence systems is cybersecurity threats. Based on hundreds of interviews of academic researchers, policy makers, business leaders and national security experts, Not With A Bug, But With A Sticker compiles the complex science of attacking AI systems and provides a front row seat to those who championed this change.
Turns out you do not need a Schwarzenegger to bring down Skynet; just stickers will do. That would make a science fiction, but fooling and attacking AI systems is already reality. What kind of attacks? What are organizations to do about this? Do things like the EU AI Act address this? What does this mean for those building AI systems? And what does this mean for you?
In this talk, Ram will provide a sweeping tour—from inside secretive government organizations to academic workshops at ski chalets to Google’s cafeteria—recounting how major AI systems remain vulnerable to the exploits of bad actors of all stripes. The steeplechase to solve this problem has already begun: Nations and organizations are aware that securing AI systems produces an indomitable advantage: the prize is not just to keep AI systems safe but also the ability to disrupt the competition’s AI systems. Grounded in real world examples of previous attacks, Ram's talk will show how adversaries can upend the reliability of otherwise robust AI systems with straightforward exploits.
Readings:
- Ram Shankar Siva Kumar, David R. O'Brien, Kendra Albert, Salome Vilojen, Law and Adversarial Machine Learning
- Ram Shankar Siva Kumar, Jeffrey Snover, David R. O'Brien, Kendra Albert, Salome Vilojen, Failure Modes in Machine Learning
- Ram Shankar Siva Kumar, Hyrum Anderson, Not With A Bug, But With A Sticker: Attacks on Machine Learning Systems and What To Do About Them
