What is your AI liability exposure right now?
Get a structured estimate of expected annual loss for your AI system, broken down across five risk categories with the assumptions made visible. The calculator is free, requires no login, and takes about three minutes to fill in.
How the calculator works
The calculator uses Fermi estimation grounded in epistemological rigour. For each relevant risk we estimate how often incidents occur and the financial impact when they do. Impact is broken into fines, legal costs, remediation, and indirect losses like brand damage. We adjust base rates from industry precedent for your specific context: jurisdiction, company size, automation level, and data sensitivity.
Your inputs drive every part of the analysis. Deployment type determines which risks apply. Jurisdiction shapes regulatory exposure (EU AI Act fines reach 7% of global turnover for prohibited practices). Company size scales absolute costs. Automation level changes the magnitude of an incident, since fewer humans in the loop means errors propagate further. Data sensitivity, especially health, biometric, or dependents' data, triggers specific regulatory regimes.
Every estimate comes back with a 90% confidence interval rather than a single number. Risk estimation is inherently uncertain, and false precision is worse than honest uncertainty. The expected value is our best guess, the range tells you the realistic spread, and the assumptions section makes the reasoning transparent.
The five risk categories
- Technical
- Model failures, data quality issues, system vulnerabilities, drift.
- Operational
- Process breakdowns, human errors, integration failures, incident response gaps.
- Legal and compliance
- Regulatory fines, contractual breaches, litigation, gaps against EU AI Act, ISO/IEC 42001, NIST AI RMF.
- Ethical and reputational
- Bias incidents, privacy concerns, transparency failures, brand damage.
- Governance
- Oversight gaps, accountability failures, documentation deficiencies, AI inventory and ownership.
What this is, and what it is not
- This is a screening estimate. It is informational, not legal advice, audit opinion, or regulatory guidance. We estimate financial loss over a one-year horizon. We do not cover existential AI risks, multi-year compounding effects, or harms that cannot be quantified financially. We assume a moderate baseline of governance maturity, and surface every assumption alongside the result.
- If you need a binding assessment, talk to qualified legal and compliance professionals. If you want a continuous, audit-ready picture of risk across every AI system in your organisation, that is what the Modulos platform is built for. This calculator is a fast, structured way to sketch the shape of your exposure before the deeper conversation.