Cybersecurity and AI Overview

Course Learning Objectives

• Understand cybersecurity challenges and the role of AI in addressing them.
• Identify and evaluate vulnerabilities in AI systems, including trapdoors and adversarial attacks.
• Explore AI's implications in offensive and defensive cybersecurity measures.
• Gain hands-on experience in AI-driven security applications and threat simulations.

Class Rules

• Students are encouraged to ask questions.
• Attendance is mandatory for final exam eligibility.

Course Assessment

• Final exam: 50%
• Midterm: 20%
• Quizzes: 10%
• Project: 20% (2-3 members per group, report and presentation required).
• Cheating and plagiarism will result in zero marks.

Introduction to AI and Security

• AI has applications that require human intelligence like decision-making and pattern recognition.

What is AI?

• Systems that perform tasks requiring human intelligence.
• Examples include decision-making and pattern recognition.

Categories of AI:

• Narrow AI: Performs specific tasks, such as spam filtering.
• General AI: Hypothetical AI with human-level intelligence.

What is Machine Learning (ML)?

• A subset of AI that trains systems to learn from data without explicit programming.

Machine Learning

• Supervised Learning: Models are trained on labeled data, e.g., malware detection.
• Unsupervised Learning: Models identify patterns without labels, e.g., anomaly detection.
• Reinforcement Learning: Systems learn by interacting with their environment and receiving feedback, e.g., AI in autonomous cybersecurity systems.

AI Enabling Security

• AI systems support security.
• Can detect anomalies, predict potential attacks, and secure systems against known and unknown threats.
• Examples include AI in intrusion detection systems and AI-driven malware analysis.

Security Enabling Better AI

• Homomorphic Encryption allows computations on encrypted data without decrypting it, ensuring privacy in AI models handling sensitive data.
• Federated Learning trains AI models across multiple devices or servers while keeping data decentralized, enhancing privacy by avoiding direct access to the data itself.

Why AI Security is Important

• AI is embedded in various sectors, from smart cities to healthcare.
• As AI takes over decision-making, attackers are more incentivized to target these systems due to the potential damage.
• AI is crucial for understanding and preparing for the evolving security landscape.

Role of AI in Cybersecurity

• Defense: Detects and prevents cyber threats through pattern recognition and real-time analysis.
• Offense: Used by attackers to automate attacks and create sophisticated malware.

Why AI is Critical in Modern Cybersecurity

• Automation provides efficiency and speed.
• Adaptive defense allows the system to learn and evolve.

AI Applications in Cybersecurity

• Intrusion Detection Systems (IDS) use anomaly detection.
• Malware detection utilizes machine learning models.
• Behavioral Analysis detects insider threats.

Challenges of AI in Security

• AI is a vulnerability.
• Examples include Adversarial Machine Learning (AML), where attackers can craft inputs to cause AI systems to make mistakes.

AI Vulnerabilities

• Data Poisoning involves manipulating training data to degrade performance.
• Adversarial Attacks introduce perturbations to fool AI models.
• Model Inversion Attacks extract sensitive data from AI models.

Trust and Bias Issues

• AI Bias can influence results.
• Lack of Transparency hinders trust in AI systems.

The Dual Use of AI

• Offensive AI: Attackers leverage AI for smart malware, spear-phishing and creating deepfakes for fraud.
• Defensive AI: AI predicts, detects and responds to security threats before harm occurs.

The importance of AI in Modern Cybersecurity

• Scalability and efficiency are crucial for managing large datasets and rapidly detecting threats.
• The evolution of cyber threats necessitates AI to adapt and evolve alongside them.

Examples of AI in Cybersecurity Tools

• Darktrace
• IBM Watson for Cybersecurity
• Vectra AI