Session 1:  Introduction to AI Security 

LIVE Session: Wednesday 23rd October @ 6pm (BST)

• What is AI Security?
  • Key concepts in AI security
•  Understanding AI Security
  • Overview of AI and machine learning
  • How security is integrated into AI systems
• Security Principles in AI
  • Confidentiality, integrity, and availability
  • Basic security principles applicable to AI

• Demo: Identifying AI Security Risks
  • Introduce a simple AI model (e.g., a basic image classifier).
  • Discuss potential security risks associated with the model.
  • Provide an example of how an AI model could be misused if not properly secured.
  • Explore the model’s behaviour and identify potential security vulnerabilities.

Session 2: Attack Vectors and Vulnerabilities in AI 

LIVE Session: Wednesday 30th October @ 6pm (BST)

• Attack Vectors in AI
  • Common attack vectors (e.g., adversarial attacks, data poisoning)
  • Real-world examples of attacks on AI systems
• Vulnerabilities in AI Models
  • Types of vulnerabilities in AI model
  • How vulnerabilities can be exploited
• Adversarial Attacks
  • What are adversarial attacks?
  • Examples and demonstrations

• Demo: Adversarial Attack on a Simple Model
  • Show how to create and apply adversarial examples to a basic AI model.
  • Discuss the impact of adversarial attacks on model performance.

Session 3: Privacy Attacks and Mitigations 

LIVE Session Wednesday 6th November @ 6pm (BST)

• Privacy Attacks on AI
  • Types of privacy attacks (e.g., membership inference, model inversion)
  • Consequences of privacy breaches in AI
• Mitigating Vulnerabilities
  • Techniques to mitigate vulnerabilities in AI models
  • Importance of robustness in AI models
• Privacy-Preserving AI
  • Techniques for preserving privacy in AI systems
  • Differential privacy, federated learning, and other methods

• Demo: Membership Inference Attack
  • Demonstrate a membership inference attack on a simple AI model.
  • Discuss the implications of such attacks on privacy.
  • Demo privacy-preserving techniques (e.g., data anonymization, differential privacy) and observe the effects on the model.

Session 4: Building Secure and Ethical AI Systems 

LIVE Session: Wednesday 13th November @ 6pm (BST)

• Building Secure AI Systems
  • Best practices for building secure AI systems
  • Design principles and security measures
• Ethical and Legal Considerations
  • Ethical issues in AI security
  • Legal frameworks and regulations
• Future Trends and Emerging Threats
  • Emerging threats in AI security
  • Future trends and developments

• Demo: Securing an AI Model
  • Walk through securing an AI model by implementing various security measures.
  • Discuss ethical considerations in the design and deployment of AI systems.