Skilled Responsible Artificial Intelligence Hacker

• Security principles applied to AI environments
• Emerging risks in AI solutions
• Security as an enabler of trust

Discussion: Identifying real initial risks

• Identity and access management in AI solutions
• Integration with access policies
• Regulatory requirements and compliance
• Overview of identity management tools

• Overview of the security incident lifecycle phases
• Detection, response, recovery, and learning
• Requirements for simulating attack and defense scenarios

Exercise: Defining an AI security incident management process

• Key components of a Security Chapter
• Roles and responsibilities in AI security
• Alignment with AI governance practices
• Discussion of a Security Chapter example

• Microsoft Purview: data governance and classification
• Microsoft Defender: protection of AI environments
• Security Copilot: incident response automation
• Overview of Purview and Defender tools

• Designing secure AI architectures
• Solution types and scenarios: Buy (SaaS), Extend & Build (PaaS / IaaS)
• Security controls by maturity level
• Discussion of security posture vs. architecture

• What is ethical hacking applied to AI?
• Differences from traditional hacking
• Key risks in AI systems (models, data, pipelines)

• OWASP AI Security & Privacy Guide
• MITRE ATLAS (Adversarial Threat Landscape for AI Systems)
• NIST AI Risk Management Framework
• How these frameworks integrate into testing methodologies

• Phase 1: Reconnaissance and Data Collection – Identification of models and datasets
• Phase 2: Analysis and Enumeration – Evaluation of architecture and weaknesses
• Phase 3: Exploitation (Adversarial Attacks): Evasion, Poisoning, Model Extraction
• Phase 4: Post-Exploitation and Reporting – Impact assessment, mitigation, and documentation of findings

• Accountability in AI testing
• Applicable regulations (GDPR, AI Act)
• Best practices for secure testing