# Scoring and Risk Library Reference

*Audience: everyone who scores, reviews, or reports on register risks.*

## How scores work

A score is **likelihood × impact**. Each is rated on five levels, so a score runs from 1 to 25 - the higher, the more urgent.

| Level | Likelihood | Impact |
|---|---|---|
| 1 | Rare | Negligible |
| 2 | Unlikely | Minor |
| 3 | Possible | Moderate |
| 4 | Likely | Major |
| 5 | Almost certain | Severe |

Scores fall into four bands. A score on a boundary belongs to the higher band.

| Band | Score range |
|---|---|
| **Low** | 1-4 |
| **Medium** | 5-9 |
| **High** | 10-16 |
| **Critical** | 17-25 |

- **Inherent risk** is the risk as it stands, before treatment.
- **Residual risk** is what's left once the treatment is in place.

A risk needs both a likelihood and an impact to be scored; until then it shows **Not scored**. Scores and bands are always calculated from the likelihood and impact, so they can't drift out of step with them.

## Field values

### Status

| Value | Meaning |
|---|---|
| **Open** | An active risk you're managing |
| **Closed** | No longer applies or fully resolved; hidden from the register by default |
| **Archived** | Kept for the record but out of day-to-day view |

### Review status

| Value | Meaning |
|---|---|
| **Draft** | Being written or scored |
| **In review** | With a reviewer |
| **Changes requested** | Sent back for changes |
| **Approved** | Signed off |

### Treatment

| Value | Meaning |
|---|---|
| **Mitigate** | Reduce likelihood or impact with controls |
| **Accept** | Live with the risk as it is, with a recorded decision |
| **Transfer** | Shift it to another party, for example through insurance or contract |
| **Avoid** | Stop the activity that causes the risk |
| **Not treated** | No treatment decided yet |

### Control role

| Value | Meaning |
|---|---|
| **In place** | An existing control already reducing the risk |
| **Planned treatment** | A control planned as part of the risk's treatment |

### Monitoring

| Value | Meaning |
|---|---|
| **Monitored by Relyance** | At least one policy is linked; its issues show on the risk |
| **Not monitorable** | The risk can't be actively monitored on the platform; track it through its controls, treatment, and reviews |
| **Not yet assessed** | No policy linked yet |

### Applies to

A risk can apply to any mix of: **Asset**, **Identity**, **Vendor**, **Service**, **Repository**, **Domain**, **AI model**, **AI agent**, **AI use case**, **MCP server**, **DSR**, and **Assessment**. A risk with nothing selected applies across the organization.

## The risk library

The library ships about 90 risk scenarios, written to be added as they are and refined afterwards:

| Domain | Scenarios | Examples |
|---|---|---|
| **Privacy** | 23 | Processing without valid consent, purpose creep, over-retention, late access requests, breach notification missed |
| **Security** | 30 | Secrets in source code, missing MFA, excessive privileges, unencrypted data, unpatched vulnerabilities, misconfigured cloud storage |
| **AI** | 23 | Prompt injection, data disclosure in AI responses, excessive agent permissions, biased training data, AI deployed without an impact assessment |
| **Third party** | 15 | No data processing agreement, unapproved sub-processors, AI providers training on your data, unlawful cross-border transfers |

Each library risk comes with:

- A title and a description of the scenario and its consequence.
- A default likelihood and impact.
- Which of confidentiality, integrity, and availability it threatens.
- A suggested treatment.
- The framework controls that address it - only controls Relyance ships, so the mapping works in every workspace.
- Public references it draws on, such as NIST SP 800-30, ISO/IEC 29100, or the MIT AI Risk Repository.

### Framework coverage

Library risks map to controls across more than 80 regulations and standards. The most widely mapped are:

- **Security** - CIS Critical Security Controls v8.1, PCI DSS, NIST SP 800-53, ISO/IEC 27001, the HIPAA Security Rule, NYDFS Part 500, NIST CSF 2.0, NIST SP 800-171, DORA, and NIS2.
- **Privacy** - GDPR, ISO/IEC 27701, CCPA, ISO/IEC 29100, ISO/IEC 27018, GLBA, the NIST Privacy Framework, UK GDPR, LGPD, and India's DPDP Act.
- **AI** - the EU AI Act, ISO/IEC 42001, the OWASP AI and ML Top 10 lists, Google SAIF, the NIST AI RMF, and MITRE ATLAS.

The **Suggested** view shows only the library risks whose controls belong to frameworks you have enabled; **All** shows the whole library.
