From the course: The OWASP Top 10 for Large Language Model (LLM) Applications: An Overview
Join today to access over 25,300 courses taught by industry experts.
What is data and model poisoning?
From the course: The OWASP Top 10 for Large Language Model (LLM) Applications: An Overview
What is data and model poisoning?
- [Instructor] In this video, we will look into one of the most dangerous threats to AI systems, data and model poisoning, which is the fourth vulnerability in 2025 OWASP Top 10 for LLM applications. OWASP says, data poisoning occurs when pre-training, fine-tuning, or embedding data is manipulated to introduce vulnerabilities, backdoors, or biases. This manipulation can compromise model security, performance, or ethical behavior, leading to harmful outputs or impaired capabilities. These attacks can be obvious or deceptive but what makes them truly dangerous is how easily they can slip in. It doesn't always require breaking into your system, but just one poison data set or compromised file can start the damage, often without immediate signs. To understand this risk better, let's walk through the LLM lifecycle and see where attackers might strike. Let's start with data collection phase, one of the earliest points where poisoning can occur. Let's say a company is building LLM to help…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.