diff --git a/index.md b/index.md index 23ed185..e230b49 100644 --- a/index.md +++ b/index.md @@ -12,12 +12,12 @@ pitch: Methodology to perform an AI System Assessment Alt text +### [OWASP AI Testing Guide Table of Contents](https://github.com/OWASP/www-project-ai-testing-guide/blob/main/Document/README.md) + As organizations increasingly adopt artificial intelligence (AI) solutions, the need for a robust framework to rigorously test AI systems for security, ethics, reliability, and compliance becomes essential. Although numerous application security testing guides exist, such as the OWASP Web Security Testing Guide (WSTG) and Mobile Security Testing Guide (MSTG), the unique risks and challenges of AI systems require specialized guidance. The OWASP AI Testing Guide aims to become the reference for identifying security, privacy, ethical, and compliance vulnerabilities inherent in AI applications. Inspired by established OWASP methodologies, the AI Testing Guide will deliver structured and practical guidance for security professionals, testers, and AI developers. This guide will be technology and industry agnostic, emphasizing applicability across various AI implementation scenarios. -### [OWASP AI Testing Guide Table of Contents](https://github.com/OWASP/www-project-ai-testing-guide/blob/main/Document/README.md) - ## Importance of AI Testing AI testing is vital because AI now underpins critical decision-making and daily operations across industries, from healthcare and finance to automotive and cybersecurity. To ensure an AI system is truly reliable, secure, accurate, and ethical, testing must go well beyond basic functionality. It needs to validate bias and fairness controls to prevent discrimination, conduct adversarial robustness checks against crafted inputs designed to fool or hijack models, and perform security and privacy assessments, such as model-extraction, data-leakage, and poisoning attack simulations. Incorporating techniques like differential privacy ensures compliance with data-protection laws while safeguarding individual records. This guide’s comprehensive approach to AI testing aims to uncover hidden risks and maintain trust in AI-driven solutions. @@ -35,5 +35,5 @@ Finally all AI models live in ever-changing environments hence ongoing monitorin The OWASP AI Testing Guide is designed to serve as a comprehensive reference for software developers, architects, data analysts, researchers, and risk officers, ensuring that AI risks are systematically addressed throughout the product development lifecycle. It outlines a robust suite of tests, ranging from data‐centric validation and fairness assessments to adversarial robustness and continuous performance monitoring, that collectively provide documented evidence of risk validation and control. By following this guidance, teams can establish the level of trust required to confidently deploy AI systems into production, with verifiable assurances that potential biases, vulnerabilities, and performance degradations have been proactively identified and mitigated. -### Start contributing [DRAFT](https://github.com/OWASP/www-project-ai-testing-guide) +### Start contributing [HERE](https://github.com/OWASP/www-project-ai-testing-guide)