From the course: Securing Generative AI: Strategies, Methodologies, Tools, and Best Practices
Unlock this course with a free trial
Join today to access over 24,800 courses taught by industry experts.
Using the OWASP ASVS to protect against insecure output handling
From the course: Securing Generative AI: Strategies, Methodologies, Tools, and Best Practices
Using the OWASP ASVS to protect against insecure output handling
- [Instructor] OWASP ASVS is very comprehensive. It's very widely used in the industry because it's a framework that is aimed to provide a standardized approach for assessing the security of web applications, and especially nowadays with web applications hosting LLM-powered applications in after all. So this framework, so the ASVS is very widely used by developers, by testers, by now AI developers and security professionals that want to ensure that the applications meet specific security requirements. So it is an open standard. It outlines the expected coverage and the level of rigor for web application security verification. And of course, it provides the basis for testing technical security controls that protect against vulnerabilities like cross-site scripting, SQL injection, server-side request forgery, cross-side request forgery, and many others. Now, the ASVS define the levels of security verifications in three different levels, level one, level two, and level three. Basically…
Contents
-
-
-
-
(Locked)
Learning objectives1m 1s
-
Defining prompt injection attacks11m 41s
-
(Locked)
Exploring real-life prompt injection attacks3m 57s
-
(Locked)
Using ChatML for OpenAI API calls to indicate to the LLM the source of prompt input10m 4s
-
(Locked)
Enforcing privilege control on LLM access to back-end systems6m 10s
-
(Locked)
Best practices around API tokens for plugins, data access, and function-level permissions3m 2s
-
(Locked)
Understanding insecure output handling attacks3m 22s
-
(Locked)
Using the OWASP ASVS to protect against insecure output handling4m 43s
-
(Locked)
-
-
-
-
-