Defending and Governing AI Systems (SF)
About This Course
In today's digital age, the use of Artificial Intelligence (AI) has proliferated across various industries, from healthcare and finance to transportation and e-commerce. However, with the increased adoption of AI, there has also been a rise in AI-targeted attacks. These attacks can range from data poisoning and model inversion to adversarial examples and model extraction. It's crucial to understand these exploits to help companies secure their AI assets effectively.
The sophistication of AI-driven threats is escalating at an alarming rate, posing a significant risk of substantial damage. Consider this: a nefarious attacker could exploit an AI model to craft adversarial exploits, cunningly designed to slip past a company’s AI-fortified security defenses. In a similar vein, an attacker could deploy a model extraction attack, effectively pilfering a company’s proprietary AI models to use them for malevolent purposes.
The repercussions of failing to safeguard AI from such exploitation or attacks are grave. Companies stand to incur hefty financial losses, suffer damage to their reputation, and potentially face legal liabilities. Furthermore, incidents related to AI security can undermine public trust in AI technologies, thereby impeding their adoption and advancement.
The Defending and Governing AI Systems in Practice course is designed to equip learners with hands-on, practical expertise in identifying and mitigating the security risks associated with AI technologies. This course will delve into the world of AI exploitation, exploring the techniques and tools used to compromise AI systems, as well as the strategies and best practices for protecting AI modules from attacks by other AI modules.
What You'll Learn
LO1. Apply AI model security fundamentals to enterprise risk contexts using security frameworks and testing methodologies
LO2. Refine system testing procedures by applying quantitative risk analysis techniques against enterprise risk management frameworks to ensure comprehensive technical assurance.
LO3 .Apply red teaming methodologies to identify vulnerabilities in generative AI models and implement technical measures to mitigate detected security risks.
LO4. Develop ethical AI governance strategies by designing and refining technical guardrails to integrate risk management into the organisational AI strategy.
LO5. Lead to implement operational content moderation and trust and safety initiatives that ensure compliance, consistency, and platform integrity while achieving organizational objectives in accordance with SOP.
LO6. Develop organizational AI risk management policies to identify, assess, and mitigate AI-related risks while ensuring safe and compliant AI use.
Entry Requirements
Professional Qualification and Other Diploma
Candidates who do not meet the above requirements will be assessed on case by case basis.