8. Risk & Threat Modelling

Systematically identifying and mitigating AI-specific threats.

Risk & Threat Modelling is the systematic process of identifying potential threats specific to AI systems and developing mitigation strategies. Unlike traditional cybersecurity threats, AI systems face unique risks such as adversarial attacks (deliberately crafted inputs designed to fool the AI), data poisoning (corrupting training data to compromise the model), model extraction (stealing intellectual property through API queries), prompt injection (manipulating AI through carefully crafted inputs), and inference attacks (deducing sensitive information from model outputs). This requires structured threat modelling frameworks adapted for AI contexts.

Traditional security approaches may not adequately address these AI-specific vulnerabilities, which can be subtle and difficult to detect. This dimension assesses whether your organisation systematically identifies and mitigates AI-specific threats.

Why It Matters

AI systems are vulnerable to unique attacks that traditional cybersecurity may not address.

Maturity Levels

BasicStandardAdvancedLeading
No risk assessment for AI systems.Basic risk mapping for AI initiatives.Structured threat modelling (e.g., STRIDE, MITRE ATT&CK) applied to AI systems.Continuous horizon scanning for emerging AI threats, with proactive mitigation strategies.

📥 Related Resources & Templates

Downloadable templates, examples, and frameworks to help you implement this dimension.

AI Risk Assessment

Premium

Comprehensive risk assessment template for evaluating AI-specific risks, likelihood, impact, and mitigation strategies.

📝 XLSX ✨ XLSX

AI Threat Modelling

Premium

Threat modelling framework and templates for identifying AI-specific security threats, attack vectors, and defenses.

📝 XLSX ✨ XLSX 📚 DOCX