AI Security: A New Cybersecurity Frontier
AI is changing both offensive and defensive cybersecurity. Learn how AI systems, connected agents and emerging attacks affect security operations, privacy, governance and business risk.
Explore prompt injection, data poisoning, MCP and MCP servers, AI-assisted attacks, model transparency and the need for human oversight.
Scroll down for further information ↓
Explore the five learning levels
Five progressive levels connect cybersecurity fundamentals with practical controls, architecture and leadership decisions. Articles and videos will be added as the library develops.
Level 1 — Foundation
- AI and the changing cyber threat landscape
- Generative AI and the security black box
- Why cybersecurity leaders need AI literacy
Level 2 — Developing
- Prompt injection and data poisoning basics
- MCP and MCP servers: tools, permissions and trust boundaries
- Privacy, access control and AI governance
Level 3 — Practitioner
- Threat modelling AI applications and agents
- Defensive controls for MCP-connected systems
- Testing AI-generated applications and workflows
Level 4 — Advanced
- Mechanistic interpretability and model behaviour
- Advanced prompt-injection and data-poisoning scenarios
- Evaluating agentic AI security architectures
Level 5 — Expert
- Enterprise AI security strategy and assurance
- AI red teaming, governance and incident response
- Business risk, human oversight and emerging threats
The five-level programme is being developed. Planned assessments contain 20 questions with an 80% pass mark. Content and assessment availability will be shown as each level is published.
Your learning, with a clear purpose
Your Pathway connects the material you study to your interests and goals. As the library grows, it will help you find relevant topics and understand the next step in your learning journey.
Knowledge you can demonstrate
Explain AI systems and threat models; recognise prompt injection and data poisoning; evaluate controls and business impact.
Each milestone will identify what you should be able to explain, evaluate and apply. Completing an article alone is not evidence of competence.