AI Cybersecurity All-in-One by Kai London
Most organisations now run AI in production and secure it with a policy document. This is the single reference that closes that gap — the threat model, the architecture, the controls and the operating discipline for enterprises that have to defend AI and defend with AI at the same time.
Available on Amazon.
What is inside
The AI threat landscape
Prompt injection, data and model poisoning, model theft and inference abuse — described as attack paths, not abstractions.
LLM and generative AI security
Input and output handling, retrieval boundaries, secrets, and where the trust boundary really sits in a RAG pipeline.
Agentic AI and tool use
Autonomy, delegation and tool calling — containing what an agent is permitted to do when it is wrong.
Adversarial machine learning
Evasion, extraction and poisoning, and the practical defences that survive contact with a production model.
AI red teaming
Structured adversarial testing of models, prompts, retrieval and agents, with findings a risk committee can read.
Zero Trust and SOC automation
Identity for AI workloads, telemetry, detection content, and where automation helps the SOC rather than flooding it.
Who it is for: CISOs and security architects standing up AI security programmes, SOC leaders adopting AI-assisted detection, and engineering teams shipping AI features into regulated enterprises.
You cannot govern what you have not architected, and you cannot defend what you have not tested.
About the author
Professor Kai London — CISSP, CISM.
An internationally recognised cybersecurity executive, board advisor and Founder & CEO of Quantum AI Systems Security LLC, writing at the convergence of AI, governance and operational resilience. Honorary Professor and Researcher at UCL.