Appearance
Acting on Quantified Risk
What this is
Quantification gives each risk threat a monetary expected loss. This page tells you what to do with that value in Modulos:
- compare it with the limits
- set priorities across the organization
- do mitigations through Controls and Evidence
- quantify again to estimate the effect
Where in Modulos
AI System → Risksto compare the quantified values with the risk limits and the AI System limitOrganization → Risk Management → Risk Overviewfor the organization totals and the risks with the highest valuesOrganization → Risk Management → Portfolioto find the AI Systems and risks that use the most of their limitsAI System → Controlsto do mitigations as Controls and attach Evidence to themAI System → Risks → [Risk] → Threat Vectors → Quantifyto quantify again after the system or its environment changes
For the roles and the rules for the limits, see Risk Operating Model.
Who can do what
- AI System Owners manage the risks and start quantification runs.
- AI System Editors add and change risks and their threat selections.
- AI System Viewers can only read, for oversight and audit.
- Risk Managers change the taxonomy, the appetite, and the limits.
Compare exposure against limits
Risk limits divide the risk appetite into budgets for categories, AI Systems, and risks. After quantification, each level shows its value against its limit:
AI System → Risksshows the value of each risk against its limit, and the total of the AI System against the AI System limit.Organization → Risk Management → Risk Overviewshows the total exposure against the risk appetite, and the use of the limits of the first five categories.Organization → Risk Management → Portfolioshows the use of each limit as a color.
A value near or above its limit is a signal. Examine the assumptions, invest in mitigation, or send the decision to the owner of that budget.
Using risk limits
Use limits as budgets for exposure:
- Below the limit: accept the risk, or let the team make the decision in the limit.
- Near the limit: examine the assumptions and do the most important mitigations first.
- Above the limit: decrease the risk, move it to a third party, stop the work, or change the scope.
Prioritize across the portfolio
The Risk Overview tab shows the use of the limits of the first five categories. It also shows the risks with the highest values in each category and the AI Systems with the highest values. The Portfolio tab shows the groups, the AI Systems, and the risks that are above their limits, down to each threat vector. See Treemap.
The Value and Risk view of the Portfolio tab compares the annual economic value of each AI System with its risk value. Use it to find the AI Systems with a Negative assessment. Their risk value is more than two times their economic value. See Value and Risk.
Start the mitigation work with the threats that have the highest expected loss.
Mitigate through Controls
Modulos does not connect risks to Controls. You do the mitigation work in the governance workflow. A new quantification of the risk threats estimates its effect:
- Quantify the baseline Get a monetary value with clear assumptions.
- Do mitigations as Controls Record the work in AI System → Controls and attach Evidence.
- Quantify again Quantify the risk threats that the mitigation changes.
- Compare with the limits Compare the new value with the risk limit and the AI System limit.
A mitigation usually decreases the rate (prevention, detection, access control, guardrails) or the damage (containment, human oversight, incident response, rollback). The division into rate and damage shows which part a mitigation changes. A new quantification shows the change in the estimated risk.
Quantification also makes the cost of mitigation a clear question: "If we spend €X, how much expected loss do we reduce?" Quantify before the investment and again after it. The difference is the estimated effect.
Re-quantify after changes
Each risk threat keeps all its quantification runs. Only the newest run of a risk threat counts in the totals, when it has the status Quantified. See How risk becomes a portfolio signal.
Quantify again in these conditions:
- you add or remove Controls for mitigation
- the system, the model, or a vendor changes
- the use increases or the deployment context changes
- an incident or a test result shows that your assumptions are incorrect
The Quantification History and Risk Value Over Time tabs show how the expected loss changes from run to run. See Run history and trends.