Overview
Explore the critical realm of AI security in this 37-minute conference talk by Johnathan Roy Azaria, a Data Scientist and Tech Lead at Imperva's threat research group. Delve into the intricate landscape of data poisoning threats and uncover tactics employed by malicious actors to sabotage AI model training, skew results, and introduce hidden vulnerabilities. Gain insights into strategies for reinforcing AI systems against covert threats, offering invaluable real-world knowledge. Cover key topics including AI risks, AI threats, data poisoning, learning from web traffic, and a comprehensive summary of the challenges and defenses in AI security.
Syllabus
- Intro
- AI Risks
- AI Treats
- Data Poisoning
- Learning from Web Traffic
- Summary
Taught by
Open Data Science