GroveAI
StrategyFree Template

AI Risk Assessment Template

A structured template for identifying, assessing, and mitigating risks across all dimensions of an AI project. Covers technical, ethical, regulatory, operational, and reputational risks with a scoring framework and mitigation planning approach.

Overview

What's included

Risk identification framework across 5 dimensions
Likelihood and impact scoring matrix
Detailed risk register with mitigation plans
Residual risk assessment methodology
Risk monitoring and review schedule
Escalation criteria and procedures
1

Risk Identification

26 itemsto complete

Risk Identification

AI system name:   Assessment date:   Assessor(s):   Risk tier (from governance framework): Tier 1 / 2 / 3

Technical Risks

  • Model accuracy below acceptable thresholds
  • Training data is unrepresentative or biased
  • Model performance degrades over time (drift)
  • System cannot handle expected throughput or latency requirements
  • Integration with existing systems fails or causes issues
  • AI outputs are not reproducible or consistent
  • Adversarial attacks or prompt injection vulnerabilities

Ethical Risks

  • Algorithmic bias affecting protected groups
  • Lack of transparency in AI decision-making
  • Unintended consequences of AI actions
  • Displacement of employee roles without proper transition
  • AI-generated content is misleading or harmful

Regulatory Risks

  • Non-compliance with GDPR or data protection laws
  • Non-compliance with the EU AI Act
  • Non-compliance with sector-specific regulations
  • Inadequate record-keeping or audit trail
  • Cross-border data transfer issues

Operational Risks

  • Single point of failure in AI infrastructure
  • Key person dependency (only one person understands the system)
  • Vendor lock-in with no exit strategy
  • Insufficient monitoring leading to undetected failures
  • Lack of rollback capability

Reputational Risks

  • Negative media coverage of AI use
  • Customer backlash due to AI errors or perceived unfairness
  • Employee distrust of AI systems
  • Competitive disadvantage from AI failures
2

Risk Scoring Matrix

Risk Scoring Matrix

Likelihood Scale

ScoreLikelihoodDescription
1RareMay occur only in exceptional circumstances
2UnlikelyCould occur but not expected
3PossibleMight occur at some time
4LikelyWill probably occur in most circumstances
5Almost certainExpected to occur in most circumstances

Impact Scale

ScoreImpactDescription
1NegligibleMinor inconvenience; no business disruption
2MinorSmall financial loss (< £10k); brief disruption
3ModerateSignificant financial loss (£10k-£100k); service degradation
4MajorLarge financial loss (£100k-£1M); extended outage; regulatory action
5SevereCritical financial loss (> £1M); existential threat; major regulatory penalty

Risk Rating

Risk Score = Likelihood x Impact

Impact 1Impact 2Impact 3Impact 4Impact 5
Likelihood 55 (Med)10 (High)15 (High)20 (Critical)25 (Critical)
Likelihood 44 (Med)8 (Med)12 (High)16 (High)20 (Critical)
Likelihood 33 (Low)6 (Med)9 (Med)12 (High)15 (High)
Likelihood 22 (Low)4 (Med)6 (Med)8 (Med)10 (High)
Likelihood 11 (Low)2 (Low)3 (Low)4 (Med)5 (Med)

Thresholds:

  • 1-4: Low — Accept and monitor
  • 5-9: Medium — Mitigate within standard processes
  • 10-15: High — Escalate; active mitigation required before deployment
  • 16-25: Critical — Escalate to governance committee; do not proceed without mitigation
3

Risk Register

Risk Register

#Risk DescriptionCategoryLikelihood (1-5)Impact (1-5)ScoreRatingMitigation StrategyOwnerStatusResidual Score
1 Technical      Open 
2 Ethical      Open 
3 Regulatory      Open 
4 Operational      Open 
5 Reputational      Open 
6        Open 
7        Open 
8        Open 

Risk Summary

  • Total risks identified:  
  • Critical risks:  
  • High risks:  
  • Medium risks:  
  • Low risks:  

Risk Review Schedule

Review TypeFrequencyNext Review DateResponsible
Risk register updateMonthly Project Lead
Full risk reassessmentQuarterly AI Governance Committee
Post-incident risk reviewAfter each incidentAs neededIncident Commander

Instructions

How to use this template

1

Identify risks across all five dimensions

Work through the risk identification checklist with your project team. Include technical, ethical, regulatory, operational, and reputational perspectives.

2

Score each risk

Assess the likelihood and impact of each risk using the scoring matrix. Be consistent in your scoring criteria across risks.

3

Define mitigation strategies

For every medium, high, and critical risk, define a specific mitigation action, owner, and timeline.

4

Assess residual risk

After defining mitigations, re-score each risk to determine the residual risk level. This tells you what risk remains even after controls are in place.

5

Establish ongoing monitoring

Set a monthly review cadence for the risk register. Risks change as projects progress and new information emerges.

Watch Out

Common mistakes to avoid

Only considering technical risks — ethical, regulatory, and reputational risks can be just as damaging.
Scoring all risks as 'medium' to avoid difficult conversations — honest scoring leads to better resource allocation.
Writing mitigations without owners — every mitigation needs a named person responsible for implementation.
Treating risk assessment as a one-time activity — risks evolve and new risks emerge throughout the project lifecycle.
Not linking the risk assessment to the governance framework — high and critical risks should trigger governance committee review.

FAQ

Frequently asked questions

Conduct an initial assessment during the planning phase, before significant investment. Update it at each major project milestone and after any significant incidents or changes.

AI risk assessments include dimensions specific to AI: model bias, data quality, explainability, algorithmic fairness, and AI-specific regulatory requirements. These are in addition to standard project risks.

Yes. Even if you did not build the AI, you are responsible for how it is used in your organisation. Assess vendor-supplied AI against the same risk framework, especially for ethical, regulatory, and reputational risks.

Escalate to the AI governance committee. Options include: accepting the risk with explicit sign-off from leadership, redesigning the solution to avoid the risk, or deciding not to proceed with the initiative.

Focus non-technical stakeholders on ethical, reputational, and business impact risks. Use plain language and concrete scenarios rather than technical jargon. Their perspective is especially valuable for identifying risks that technical teams may overlook.

Need a custom AI template?

Our team can build tailored templates for your specific business needs. Book a free strategy call.