Downloads¶
Downloadable resources for offline reading, training, and distribution. These materials complement the online reference and can be used independently.
Position Papers¶
Why Containment Beats Evaluation¶
A position paper arguing that runtime containment, constraining what AI systems can do, is more reliable than evaluation-based approaches that try to determine what AI systems will do.
Covers the limitations of pre-deployment evaluation, the case for runtime enforcement, and how containment-first thinking changes control architecture.
See also: Why Containment Beats Evaluation (online version)
AI Security Maturity Model¶
A position paper on how organisational maturity affects AI security posture. Covers what Level 1 and Level 5 organisations look like across four adoption tracks (AI you build, platforms you build on, copilots, AI coding tools), the balance between AI and non-AI security solutions, data management as a gate, and why the real gap is operational discipline rather than capability.
Grounded in established maturity models (CMMI, NIST CSF, C2M2, ISO 42001) and current industry data, with an honest assessment of which commonly cited statistics are reliable.
See also: Maturity Levels (online version)
Framework Reference¶
MASO Framework (Full PDF)¶
The complete Multi-Agent Security Operations (MASO) framework as a single PDF document. Includes all 7 control domains (128 controls), implementation tiers, red team playbook, incident tracker, integration guide, and worked examples.
See also: Multi-Agent Security Operations (MASO) (online reference)
Training Materials¶
MASO Practitioner Training v.1¶
Comprehensive training material for practitioners implementing Multi-Agent Security Operations (MASO). Covers the MASO framework, control domains, implementation tiers, and practical guidance for securing multi-agent AI systems.
See also: Multi-Agent Security Operations (MASO) (online reference)
All materials are provided under the MIT License.