Author: Black Hat - Bewertung: 1x - Views:7
As AI agents become integral to finance, healthcare, and customer service, addressing their unique privacy and security challenges is crucial. This session explores key vulnerabilities in multi-agent AI systems, including adversarial attacks, prompt injections, and social engineering risks. Through real-world scenarios and case studies, we'll examine how these agents operate autonomously, learn, and adapt—and how these features can be exploited. Attendees will gain actionable insights on building robust defenses, from privilege management and LLM "judge" mechanisms to strict communication protocols, empowering them to fortify AI deployments against evolving threats.
By:
Narayana Pappu | CEO, Zendata
Rubens Zimbres | Senior AI Engineer
Full Abstract and Presentation Materials:
https://www.blackhat.com/eu-24/briefings/schedule/#mind-the-data-gap-privacy-challenges-in-autonomous-ai-agents-43069