The Darker Side of AI

Startup pitch AI Application AI REGULATION AI Technology COMPUTER VISION CONVERSATIONAL AI CYBERSECURITY DECISION AUTOMATION DEEP LEARNING EXPLAINABLE FACE RECOGNITION MODEL DRIFT MODEL MONITORING MODEL VALIDATION NATURAL LANGUAGE PROCESSING (NLP) SECURITY TRUST

08/02/2024 | 11h00 - 11h30 | Startup pitch stage 2

Information

We are all adopting artificial intelligence to enable the processing and analysis of vast amounts of data at a speed and scale that is beyond human capacity, allowing faster and more accurate decisions that spur innovation.

But attackers are taking advantage of this opportunity and Large Language Models (LLM) often lie and make mistakes. The question is not if, but when and how. From a small detail to well known facts, AI systems are only as good as the data they are trained on and the algorithms that power them.

DeepKeep’s CTO, Yossi Altevet, will describe attacks with GenAI and on GenAI that lead to theft, terrorism and illegal immigration. Yossi will explain how they breach LLM guardrails to extract data, or use visual adversarial noise to evade facial recognition platforms.