arckit-cli 6.1.1__py3-none-any.whl

arckit_cli-6.1.1.data/data/share/arckit/.arckit/templates/uk-gov-ai-playbook-template.md

@@ -0,0 +1,942 @@
+# UK Government AI Playbook Assessment
+
+> **Template Origin**: Official | **ArcKit Version**: [VERSION] | **Command**: `/arckit:ai-playbook`
+
+## Document Control
+
+<!-- DOC-CONTROL-HEADER -->
+<!-- Resolved at command-execution time to _partials/document-control-uk.md or _partials/document-control-uae.md based on plugin userConfig classification_scheme + governance_framework. See _partials/RENDERING.md (when present). -->
+
+## Revision History
+
+| Version | Date | Author | Changes | Approved By | Approval Date |
+|---------|------|--------|---------|-------------|---------------|
+| [VERSION] | [DATE] | ArcKit AI | Initial creation from `/arckit:ai-playbook` command | PENDING | PENDING |
+
+---
+
+## Executive Summary
+
+**Overall Compliance**: [Score/10] principles compliant
+
+**Risk Assessment**:
+
+- [ ] HIGH-RISK (fully automated decisions affecting rights, safety, health)
+- [ ] MEDIUM-RISK (significant impact with human oversight)
+- [ ] LOW-RISK (productivity tools, administrative tasks)
+
+**Status**:
+
+- ✅ COMPLIANT (9-10 principles met)
+- ⚠️ PARTIALLY COMPLIANT (7-8 principles met)
+- ❌ NON-COMPLIANT (< 7 principles met)
+
+**Critical Issues**: [Number] blocking issues
+**Go/No-Go Decision**: [ ] APPROVED / [ ] APPROVED WITH CONDITIONS / [ ] REJECTED
+
+---
+
+## AI System Overview
+
+### What is the AI System?
+
+**System Name**: [Name]
+
+**Purpose**: [What problem does it solve?]
+
+**Type of AI**:
+
+- [ ] Generative AI (e.g., Large Language Models, image generation)
+- [ ] Predictive AI (e.g., risk scoring, forecasting)
+- [ ] Computer Vision (e.g., image recognition, object detection)
+- [ ] Natural Language Processing (e.g., sentiment analysis, translation)
+- [ ] Recommendation System
+- [ ] Robotic Process Automation with AI
+- [ ] Other: [Specify]
+
+**Use Case**:
+[Describe how the AI will be used in government operations]
+
+**Users**:
+
+- Internal users: [Who in government uses it?]
+- External users: [Citizens, businesses affected?]
+- Affected population: [Who is impacted by decisions?]
+
+**Decision Authority**:
+
+- [ ] AI makes recommendations, humans decide
+- [ ] AI makes decisions with human review
+- [ ] AI makes autonomous decisions (HIGH-RISK - justify carefully)
+
+---
+
+## The 10 Core Principles Assessment
+
+### 1. Understanding AI
+
+**Principle**: Organizations must comprehend what AI is, its capabilities, and limitations.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Team understands AI is not sentient or reasoning
+- [ ] AI limitations documented (hallucinations, biases, edge cases)
+- [ ] Use cases appropriate for AI capabilities
+- [ ] Realistic expectations set with stakeholders
+- [ ] Technical constraints understood (data quality, compute, latency)
+
+**AI Limitations Documented**:
+| Limitation | Impact | Mitigation |
+|------------|--------|------------|
+| [e.g., Hallucinations in LLM] | [May generate false information] | [Human review of all outputs] |
+| [e.g., Bias in training data] | [May discriminate against groups] | [Fairness testing, bias mitigation] |
+
+**Findings**:
+[Describe team's understanding of AI capabilities and limitations]
+
+**Gaps**:
+[List misunderstandings or unrealistic expectations]
+
+**Score**: ___/10
+
+---
+
+### 2. Lawful and Ethical Use
+
+**Principle**: AI deployment must comply with legal requirements and ethical standards.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Legal review completed (data protection, equality, human rights)
+- [ ] Data Protection Impact Assessment (DPIA) completed
+- [ ] Equality Impact Assessment (EqIA) completed
+- [ ] Human Rights assessment completed
+- [ ] UK GDPR compliance verified
+- [ ] Equality Act 2010 compliance verified
+- [ ] Public Sector Equality Duty considered
+- [ ] Data Ethics Framework applied
+- [ ] Early engagement with legal/compliance teams
+
+**Legal and Ethical Checks**:
+| Check | Status | Issues Found | Resolution |
+|-------|--------|--------------|------------|
+| DPIA | ✅ / 🔄 / ❌ | [Issues] | [Actions] |
+| EqIA | ✅ / 🔄 / ❌ | [Issues] | [Actions] |
+| Human Rights | ✅ / 🔄 / ❌ | [Issues] | [Actions] |
+| UK GDPR | ✅ / 🔄 / ❌ | [Issues] | [Actions] |
+| Equality Act | ✅ / 🔄 / ❌ | [Issues] | [Actions] |
+
+**Data Protection**:
+
+- Legal basis for processing: [Consent / Legitimate Interest / Public Task / etc.]
+- Special category data: [ ] Yes / [ ] No (if yes, justify)
+- Data retention period: [X months/years]
+- Right to object: [ ] Enabled / [ ] Not applicable
+
+**Findings**:
+[Describe legal and ethical compliance]
+
+**Gaps**:
+[List legal or ethical risks]
+
+**Score**: ___/10
+
+---
+
+### 3. Security
+
+**Principle**: Systems must be secure and resilient to cyber attacks, including AI-specific threats.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Cyber security assessment completed
+- [ ] NCSC guidance followed
+- [ ] AI-specific threats assessed (prompt injection, data poisoning, model theft)
+- [ ] Security controls implemented
+- [ ] Penetration testing completed
+- [ ] Red teaming conducted (for high-risk AI)
+- [ ] Incident response plan includes AI-specific scenarios
+- [ ] Supply chain security verified (third-party AI services)
+
+**AI-Specific Security Threats**:
+| Threat | Risk Level | Mitigation |
+|--------|------------|------------|
+| Prompt Injection | [H/M/L] | [Input sanitization, content filtering] |
+| Data Poisoning | [H/M/L] | [Data validation, anomaly detection] |
+| Model Theft | [H/M/L] | [Access controls, API rate limiting] |
+| Adversarial Attacks | [H/M/L] | [Robustness testing, input validation] |
+| Model Inversion | [H/M/L] | [Differential privacy, access controls] |
+
+**Security Controls**:
+
+- [ ] Input validation and sanitization
+- [ ] Output content filtering (for generative AI)
+- [ ] Rate limiting on API endpoints
+- [ ] Access controls (RBAC/ABAC)
+- [ ] Audit logging of all AI interactions
+- [ ] Encryption at rest and in transit
+- [ ] Secure model storage and versioning
+- [ ] Regular security updates and patching
+
+**Findings**:
+[Describe security implementation]
+
+**Gaps**:
+[List security vulnerabilities]
+
+**Score**: ___/10
+
+---
+
+### 4. Human Control
+
+**Principle**: Meaningful human oversight must exist at appropriate stages.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Human-in-the-loop design implemented
+- [ ] Decision points require human approval
+- [ ] Override capability exists for humans
+- [ ] Escalation process documented
+- [ ] Human review frequency defined
+- [ ] Staff training on AI system limitations
+- [ ] Clear responsibilities assigned
+
+**Human Oversight Model**:
+
+- [ ] **Human-in-the-loop**: Human reviews EVERY decision before implementation
+- [ ] **Human-on-the-loop**: Human reviews decisions periodically/randomly
+- [ ] **Human-in-command**: Human can override AI decisions at any time
+- [ ] **Fully automated**: AI acts autonomously (HIGH-RISK - justify!)
+
+**Human Review Requirements**:
+| Decision Type | Review Requirement | Reviewer Role | Escalation Path |
+|---------------|-------------------|---------------|-----------------|
+| [High-impact decisions] | Every decision | [Senior officer] | [SRO] |
+| [Medium-impact] | Random sample (10%) | [Team lead] | [Senior officer] |
+| [Low-impact] | Audit trail only | [Automated monitoring] | [Team lead] |
+
+**For High-Risk AI** (affecting health, safety, fundamental rights):
+
+- [ ] MUST have human-in-the-loop (review every decision)
+- [ ] Humans trained on AI limitations and biases
+- [ ] Override process tested and documented
+- [ ] Decision rationale explainable to affected persons
+
+**Findings**:
+[Describe human oversight mechanisms]
+
+**Gaps**:
+[List areas lacking sufficient human control]
+
+**Score**: ___/10
+
+---
+
+### 5. Lifecycle Management
+
+**Principle**: Understand complete product lifecycle from selection to decommissioning.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Lifecycle plan documented
+- [ ] Supplier selection criteria defined
+- [ ] Contract includes performance metrics and SLAs
+- [ ] Model versioning and change management
+- [ ] Monitoring and performance tracking
+- [ ] Retraining schedule defined
+- [ ] Model drift detection implemented
+- [ ] Decommissioning plan (data deletion, model retirement)
+- [ ] Handover documentation prepared
+
+**Lifecycle Stages**:
+
+**1. Selection and Procurement**:
+
+- [ ] Requirements defined (see ARC-{PROJECT_ID}-REQ-v*.md)
+- [ ] Build vs buy decision documented
+- [ ] Supplier evaluation completed (see ARC-*-EVAL-*.md)
+- [ ] Contract includes AI-specific terms (performance, bias, retraining)
+
+**2. Development and Testing**:
+
+- [ ] Training data provenance documented
+- [ ] Bias testing completed
+- [ ] Performance benchmarks established
+- [ ] User acceptance testing (UAT) with real users
+- [ ] Accessibility testing completed
+
+**3. Deployment**:
+
+- [ ] Phased rollout plan (pilot, beta, full deployment)
+- [ ] Monitoring dashboards configured
+- [ ] Alert thresholds defined
+- [ ] Incident response procedures ready
+
+**4. Operation and Maintenance**:
+
+- [ ] Ongoing performance monitoring
+- [ ] Model drift detection (monthly checks)
+- [ ] Retraining schedule (e.g., quarterly with new data)
+- [ ] User feedback collection mechanism
+- [ ] Regular fairness and bias audits
+
+**5. Decommissioning**:
+
+- [ ] Data deletion procedure defined
+- [ ] Model archive or deletion
+- [ ] User notification plan
+- [ ] Alternative process for affected users
+- [ ] Lessons learned documentation
+
+**Model Monitoring Metrics**:
+| Metric | Baseline | Threshold | Current | Action if Exceeded |
+|--------|----------|-----------|---------|-------------------|
+| Accuracy | [%] | [%] | [%] | [Retrain model] |
+| False Positive Rate | [%] | [%] | [%] | [Tune threshold] |
+| Fairness (demographic parity) | [±%] | [±%] | [±%] | [Bias mitigation] |
+| Latency | [ms] | [ms] | [ms] | [Scale infrastructure] |
+
+**Findings**:
+[Describe lifecycle management approach]
+
+**Gaps**:
+[List missing lifecycle processes]
+
+**Score**: ___/10
+
+---
+
+### 6. Right Tool Selection
+
+**Principle**: AI should only be deployed when it genuinely solves problems better than alternatives.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Problem clearly defined
+- [ ] Alternative solutions considered (non-AI, simpler AI)
+- [ ] Cost-benefit analysis completed
+- [ ] AI adds genuine value over alternatives
+- [ ] Use case appropriate for AI (not using AI for sake of it)
+- [ ] Success metrics defined
+- [ ] Pilot/proof-of-concept completed
+
+**Alternatives Considered**:
+| Solution | Pros | Cons | Why Not Chosen |
+|----------|------|------|----------------|
+| Manual process | [Accurate] | [Slow, expensive] | [Cannot scale to demand] |
+| Rule-based system | [Explainable] | [Inflexible] | [Too many edge cases] |
+| Simple ML (not AI) | [Faster] | [Less accurate] | [Accuracy critical for use case] |
+| AI (selected) | [Accurate, scalable] | [Less explainable, bias risk] | [Best fit with mitigation] |
+
+**Use Case Appropriateness**:
+
+- [ ] Problem is well-suited to AI capabilities
+- [ ] Sufficient quality training data available
+- [ ] Success can be measured objectively
+- [ ] Acceptable trade-offs (e.g., explainability vs accuracy)
+- [ ] NOT using AI just because it's trendy
+
+**Inappropriate Use Cases to Avoid**:
+
+- [ ] Fully automated decisions on life-changing matters (without justification)
+- [ ] High-stakes decisions with insufficient training data
+- [ ] Use cases requiring 100% accuracy where AI can't achieve it
+- [ ] Situations where simpler solutions are adequate
+
+**Success Metrics**:
+| Metric | Baseline | Target | Current | Status |
+|--------|----------|--------|---------|--------|
+| [Accuracy] | [%] | [%] | [%] | ✅ / ⚠️ / ❌ |
+| [Processing time] | [X hours] | [X minutes] | [X minutes] | ✅ / ⚠️ / ❌ |
+| [Cost per transaction] | [£X] | [£Y] | [£Z] | ✅ / ⚠️ / ❌ |
+| [User satisfaction] | [X/10] | [Y/10] | [Z/10] | ✅ / ⚠️ / ❌ |
+
+**Findings**:
+[Describe rationale for AI selection]
+
+**Gaps**:
+[List concerns about appropriateness]
+
+**Score**: ___/10
+
+---
+
+### 7. Collaboration
+
+**Principle**: Engage across government and with external stakeholders.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Cross-government collaboration (GDS, CDDO, AI Standards Hub)
+- [ ] Academia partnerships
+- [ ] Industry engagement
+- [ ] Civil society consultation
+- [ ] User community involvement
+- [ ] Sharing lessons learned
+- [ ] Contributing to government AI community
+
+**Collaboration Activities**:
+| Stakeholder | Engagement Type | Purpose | Outcome |
+|-------------|-----------------|---------|---------|
+| [GDS] | [Workshop] | [Best practices] | [Adopted design patterns] |
+| [University X] | [Research partnership] | [Bias mitigation] | [Improved fairness metrics] |
+| [Civil society org] | [Consultation] | [Rights impact] | [Enhanced safeguards] |
+| [Other dept] | [Knowledge sharing] | [Similar use case] | [Reused components] |
+
+**Knowledge Sharing**:
+
+- [ ] Documented lessons learned
+- [ ] Presented at cross-government forums
+- [ ] Published case studies (where appropriate)
+- [ ] Contributed to reusable components
+
+**Findings**:
+[Describe collaboration efforts]
+
+**Gaps**:
+[List missed collaboration opportunities]
+
+**Score**: ___/10
+
+---
+
+### 8. Commercial Partnership
+
+**Principle**: Early engagement with commercial colleagues on responsible AI expectations.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Procurement team engaged early
+- [ ] AI requirements in contract (performance, explainability, bias)
+- [ ] Supplier responsible AI commitments documented
+- [ ] Audit rights included in contract
+- [ ] Data rights and ownership clear
+- [ ] Exit strategy defined (data portability, model ownership)
+- [ ] Performance metrics and SLAs
+- [ ] Liability and indemnity clauses for AI failures
+
+**Contract Requirements for AI**:
+
+- [ ] **Performance metrics**: Accuracy, latency, uptime SLAs
+- [ ] **Explainability**: Supplier must explain how AI works
+- [ ] **Bias audits**: Regular fairness testing required
+- [ ] **Retraining**: Frequency and process for model updates
+- [ ] **Data rights**: Government owns training data and outputs
+- [ ] **Audit rights**: Government can audit AI system
+- [ ] **Security**: Cyber security standards compliance
+- [ ] **Liability**: Clear accountability for AI failures
+- [ ] **Exit**: Data portability, model handover, decommissioning
+
+**Supplier Responsible AI Commitments**:
+| Commitment | Contractual? | Verification Method |
+|------------|--------------|---------------------|
+| [Bias testing quarterly] | ✅ Yes | [Audit reports] |
+| [Explainability documentation] | ✅ Yes | [Technical docs] |
+| [Data security standards] | ✅ Yes | [Cyber Essentials Plus cert] |
+| [Human oversight capability] | ✅ Yes | [Override mechanism tested] |
+
+**Findings**:
+[Describe commercial partnership approach]
+
+**Gaps**:
+[List missing contract terms or supplier commitments]
+
+**Score**: ___/10
+
+---
+
+### 9. Skills and Expertise
+
+**Principle**: Teams need appropriate technical, ethical, and domain expertise.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] AI/ML technical expertise on team
+- [ ] Data science capability
+- [ ] Ethical AI expertise or access
+- [ ] Domain expertise (understanding of problem domain)
+- [ ] User research capability
+- [ ] Legal/compliance expertise
+- [ ] Cyber security expertise
+- [ ] Training provided to all team members
+
+**Team Composition**:
+| Role | Filled? | Name/Team | Expertise Level |
+|------|---------|-----------|-----------------|
+| AI/ML Specialist | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| Data Scientist | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| Ethics Advisor | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| Domain Expert | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| User Researcher | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| Legal/Compliance | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| Security Specialist | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+| Product Manager | ✅ / ❌ | [Name] | [Junior/Mid/Senior] |
+
+**Training Provided**:
+
+- [ ] AI fundamentals for all team members
+- [ ] Ethical AI considerations
+- [ ] Bias recognition and mitigation
+- [ ] AI system limitations
+- [ ] User research with AI systems
+- [ ] Incident response for AI failures
+
+**Findings**:
+[Describe team capabilities]
+
+**Gaps**:
+[List missing skills or expertise]
+
+**Score**: ___/10
+
+---
+
+### 10. Organizational Alignment
+
+**Principle**: AI use must align with organizational policies, governance, and assurance.
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] AI governance board approval obtained
+- [ ] AI strategy alignment documented
+- [ ] Organizational AI principles followed
+- [ ] Risk management process followed
+- [ ] Assurance team engaged throughout
+- [ ] Escalation process defined and followed
+- [ ] Senior Responsible Owner (SRO) assigned
+- [ ] Regular governance reviews scheduled
+
+**Governance Structure**:
+
+- **AI Governance Board**: [Name of board]
+- **Senior Responsible Owner**: [Name, title]
+- **Product Owner**: [Name, title]
+- **Assurance Lead**: [Name, title]
+
+**Governance Checkpoints**:
+| Phase | Review Required | Status | Date | Outcome |
+|-------|-----------------|--------|------|---------|
+| Concept | AI Governance Board | ✅ / 🔄 / ⏳ | [Date] | [Approved/Conditions] |
+| Design | Technical Review | ✅ / 🔄 / ⏳ | [Date] | [Approved/Conditions] |
+| Pre-Deployment | Security & Ethics Review | ✅ / 🔄 / ⏳ | [Date] | [Approved/Conditions] |
+| Post-Deployment | Performance Review | ✅ / 🔄 / ⏳ | [Date] | [Approved/Conditions] |
+
+**Organizational AI Principles** (if defined):
+
+- [ ] Aligns with department's AI principles
+- [ ] Aligns with cross-government AI principles
+- [ ] No conflicts with organizational values
+
+**Findings**:
+[Describe organizational alignment]
+
+**Gaps**:
+[List governance or alignment issues]
+
+**Score**: ___/10
+
+---
+
+## Six Ethical Themes Assessment
+
+### 1. Safety, Security, and Robustness
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Safety testing completed (no harmful outputs)
+- [ ] Robustness testing (handles edge cases)
+- [ ] Security controls implemented (see Principle 3)
+- [ ] Fail-safe mechanisms in place
+- [ ] Incident response plan tested
+
+**Safety Measures**:
+| Risk | Safeguard | Testing | Status |
+|------|-----------|---------|--------|
+| [Harmful content generation] | [Content filtering] | [Red team testing] | ✅ / ⚠️ / ❌ |
+| [Biased outcomes] | [Fairness testing] | [Demographic analysis] | ✅ / ⚠️ / ❌ |
+| [System failures] | [Graceful degradation] | [Chaos engineering] | ✅ / ⚠️ / ❌ |
+
+**Score**: ___/10
+
+---
+
+### 2. Transparency and Explainability
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Algorithmic Transparency Recording Standard (ATRS) completed
+- [ ] System documented publicly (where appropriate)
+- [ ] Decision explanations available to affected persons
+- [ ] Model card or factsheet published
+- [ ] Privacy notice includes AI use
+
+**ATRS Compliance**:
+
+- **ATRS URL**: [Link to published ATRS entry]
+- **Publication Date**: [Date]
+- **Last Updated**: [Date]
+
+**Explainability Level**:
+
+- [ ] **Full explainability**: Can explain why each decision was made
+- [ ] **Partial explainability**: Can explain general logic, not individual decisions
+- [ ] **Black box**: Cannot explain decisions (must justify if high-risk)
+
+**Public Communication**:
+
+- [ ] Citizens informed AI is being used
+- [ ] How to request human review explained
+- [ ] Complaint mechanism published
+
+**Score**: ___/10
+
+---
+
+### 3. Fairness, Bias, and Discrimination
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Bias assessment completed
+- [ ] Training data reviewed for bias
+- [ ] Fairness metrics calculated
+- [ ] Protected characteristics analysis
+- [ ] Bias mitigation techniques applied
+- [ ] Ongoing monitoring for bias drift
+
+**Fairness Testing**:
+| Protected Characteristic | Metric | Baseline | Threshold | Current | Status |
+|-------------------------|--------|----------|-----------|---------|--------|
+| Gender | Demographic parity | ±5% | ±10% | [%] | ✅ / ⚠️ / ❌ |
+| Ethnicity | Equal opportunity | ±5% | ±10% | [%] | ✅ / ⚠️ / ❌ |
+| Age | Equalized odds | ±5% | ±10% | [%] | ✅ / ⚠️ / ❌ |
+| Disability | Calibration | ±5% | ±10% | [%] | ✅ / ⚠️ / ❌ |
+
+**Bias Mitigation**:
+
+- [ ] Diverse training data sourced
+- [ ] Data augmentation for underrepresented groups
+- [ ] Algorithmic fairness techniques applied
+- [ ] Regular fairness audits scheduled
+
+**Score**: ___/10
+
+---
+
+### 4. Accountability and Responsibility
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Clear ownership assigned (SRO, Product Owner)
+- [ ] Decision-making process documented
+- [ ] Audit trail of all AI decisions
+- [ ] Incident response procedures
+- [ ] Accountability for errors defined
+
+**Accountability Structure**:
+
+- **Senior Responsible Owner**: [Name] - Strategic oversight
+- **Product Owner**: [Name] - Day-to-day operation
+- **Technical Lead**: [Name] - Technical implementation
+- **Ethics Lead**: [Name] - Ethical oversight
+
+**Incident Response**:
+
+- [ ] Process for reporting AI errors
+- [ ] Root cause analysis procedure
+- [ ] Corrective action tracking
+- [ ] Learning and improvement loop
+
+**Score**: ___/10
+
+---
+
+### 5. Contestability and Redress
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Right to contest AI decisions enabled
+- [ ] Human review process for contested decisions
+- [ ] Appeal mechanism documented and accessible
+- [ ] Redress process for those harmed
+- [ ] Response times defined (e.g., 28 days)
+
+**Contestability Process**:
+
+1. **How users can contest**: [Email form, online portal, phone]
+2. **Information required**: [What users must provide]
+3. **Review timeline**: [X working days]
+4. **Human reviewer**: [Role/team]
+5. **Appeal if unsatisfied**: [Next level escalation]
+6. **Redress options**: [Correction, compensation, apology]
+
+**Testing**:
+
+- [ ] Contestability process tested with real users
+- [ ] Response times meet targets
+- [ ] Users satisfied with process
+
+**Score**: ___/10
+
+---
+
+### 6. Societal Wellbeing and Public Good
+
+**Compliance Status**: [ ] COMPLIANT / [ ] PARTIAL / [ ] NON-COMPLIANT
+
+**Evidence**:
+
+- [ ] Positive societal impact assessment
+- [ ] Environmental impact considered (carbon footprint of AI)
+- [ ] Benefits distributed fairly across society
+- [ ] Negative impacts mitigated
+- [ ] Alignment with public values
+
+**Societal Impact**:
+| Impact | Positive/Negative | Magnitude | Mitigation/Enhancement |
+|--------|-------------------|-----------|------------------------|
+| [Improved service access] | Positive | High | [Proactively promote to underserved] |
+| [Job displacement] | Negative | Medium | [Reskilling programs, redeployment] |
+| [Environmental (compute)] | Negative | Low | [Efficient models, renewable energy] |
+
+**Public Good**:
+
+- [ ] Solves real problem for citizens
+- [ ] Accessible to all (not just tech-savvy)
+- [ ] Maintains human dignity and autonomy
+- [ ] Strengthens democratic values
+
+**Score**: ___/10
+
+---
+
+## Overall Assessment Summary
+
+### Compliance Scorecard
+
+| Assessment Area | Score /10 | Status |
+|-----------------|-----------|--------|
+| **10 Core Principles** | | |
+| 1. Understanding AI | __ | ✅ / ⚠️ / ❌ |
+| 2. Lawful and Ethical Use | __ | ✅ / ⚠️ / ❌ |
+| 3. Security | __ | ✅ / ⚠️ / ❌ |
+| 4. Human Control | __ | ✅ / ⚠️ / ❌ |
+| 5. Lifecycle Management | __ | ✅ / ⚠️ / ❌ |
+| 6. Right Tool Selection | __ | ✅ / ⚠️ / ❌ |
+| 7. Collaboration | __ | ✅ / ⚠️ / ❌ |
+| 8. Commercial Partnership | __ | ✅ / ⚠️ / ❌ |
+| 9. Skills and Expertise | __ | ✅ / ⚠️ / ❌ |
+| 10. Organizational Alignment | __ | ✅ / ⚠️ / ❌ |
+| **Principles Subtotal** | **__/100** | |
+| | | |
+| **6 Ethical Themes** | | |
+| 1. Safety, Security, Robustness | __ | ✅ / ⚠️ / ❌ |
+| 2. Transparency, Explainability | __ | ✅ / ⚠️ / ❌ |
+| 3. Fairness, Bias, Discrimination | __ | ✅ / ⚠️ / ❌ |
+| 4. Accountability, Responsibility | __ | ✅ / ⚠️ / ❌ |
+| 5. Contestability, Redress | __ | ✅ / ⚠️ / ❌ |
+| 6. Societal Wellbeing, Public Good | __ | ✅ / ⚠️ / ❌ |
+| **Ethics Subtotal** | **__/60** | |
+| | | |
+| **TOTAL SCORE** | **__/160** | |
+
+**Percentage Score**: ___%
+
+### Compliance Levels
+
+- **90-100%** (144-160 points): Excellent - Exemplary responsible AI
+- **75-89%** (120-143 points): Good - Compliant with minor improvements needed
+- **60-74%** (96-119 points): Adequate - Significant gaps to address
+- **< 60%** (< 96 points): Poor - Major compliance issues
+
+### Risk-Based Decision
+
+**For HIGH-RISK AI** (fully automated decisions affecting rights/safety/health):
+
+- [ ] MUST score ≥ 90% to proceed
+- [ ] ALL principles must be met (no ❌ allowed)
+- [ ] Human-in-the-loop REQUIRED
+- [ ] ATRS publication MANDATORY
+- [ ] Regular audits (quarterly minimum)
+
+**For MEDIUM-RISK AI**:
+
+- [ ] SHOULD score ≥ 75% to proceed
+- [ ] Critical principles must be met (2, 3, 4)
+- [ ] Human oversight required
+- [ ] Annual audits
+
+**For LOW-RISK AI**:
+
+- [ ] SHOULD score ≥ 60% to proceed
+- [ ] Basic safeguards in place
+- [ ] Periodic review (annual)
+
+### Critical Issues (Blocking)
+
+1. [Issue description]
+2. [Issue description]
+3. [Issue description]
+
+### Recommendations
+
+#### High Priority (Address before deployment)
+
+1. [Recommendation]
+2. [Recommendation]
+
+#### Medium Priority (Address within 3 months)
+
+1. [Recommendation]
+2. [Recommendation]
+
+#### Low Priority (Continuous improvement)
+
+1. [Recommendation]
+2. [Recommendation]
+
+---
+
+## Action Plan
+
+| Action | Principle/Theme | Owner | Due Date | Status |
+|--------|----------------|-------|----------|--------|
+| [Action item] | [Principle #] | [Name] | [Date] | 🔄 / ⏳ / ✅ |
+
+---
+
+## Mandatory Documentation
+
+### Required Assessments (attach or link)
+
+- [ ] **ATRS** (Algorithmic Transparency Recording Standard): [Link]
+- [ ] **DPIA** (Data Protection Impact Assessment): [Link]
+- [ ] **EqIA** (Equality Impact Assessment): [Link]
+- [ ] **Human Rights Assessment**: [Link]
+- [ ] **Security Risk Assessment**: [Link]
+- [ ] **Bias Audit Report**: [Link]
+- [ ] **User Research Report**: [Link]
+
+### Governance Approvals
+
+- [ ] AI Governance Board approval: [Date]
+- [ ] Senior Responsible Owner sign-off: [Date]
+- [ ] Legal review: [Date]
+- [ ] Security review: [Date]
+- [ ] Ethics review: [Date]
+
+---
+
+## Go/No-Go Decision
+
+**Decision**: [ ] APPROVED / [ ] APPROVED WITH CONDITIONS / [ ] REJECTED
+
+**Conditions for Approval** (if applicable):
+
+1. [Condition 1]
+2. [Condition 2]
+3. [Condition 3]
+
+**Deployment Approval**: [ ] Yes / [ ] No
+
+**Ongoing Monitoring Required**:
+
+- [ ] Weekly performance reviews (first month)
+- [ ] Monthly bias audits
+- [ ] Quarterly governance reviews
+- [ ] Annual comprehensive reassessment
+
+---
+
+## Sign-Off
+
+**Assessed By**: [Name, Role]
+**Date**: [Date]
+
+**Senior Responsible Owner**:
+Name: ________________
+Date: ________________
+Signature: ________________
+
+**AI Governance Board Chair**:
+Name: ________________
+Date: ________________
+Signature: ________________
+
+---
+
+## Review Schedule
+
+**Next Review**: [Date]
+**Review Frequency**:
+
+- [ ] Monthly (high-risk)
+- [ ] Quarterly (medium-risk)
+- [ ] Annually (low-risk)
+
+---
+
+**Template Version**: 1.0
+**Last Updated**: 2025-10-14
+**Source**: https://www.gov.uk/government/publications/ai-playbook-for-the-uk-government
+
+## External References
+
+> This section provides traceability from generated content back to source documents.
+> Follow citation instructions in the project's citation reference guide.
+
+### Document Register
+
+| Doc ID | Filename | Type | Source Location | Description |
+|--------|----------|------|-----------------|-------------|
+| *None provided* | — | — | — | — |
+
+### Citations
+
+| Citation ID | Doc ID | Page/Section | Category | Quoted Passage |
+|-------------|--------|--------------|----------|----------------|
+| — | — | — | — | — |
+
+### Unreferenced Documents
+
+| Filename | Source Location | Reason |
+|----------|-----------------|--------|
+| — | — | — |
+
+---
+
+**Generated by**: ArcKit `/arckit:ai-playbook` command
+**Generated on**: [DATE]
+**ArcKit Version**: [VERSION]
+**Project**: [PROJECT_NAME]
+**Model**: [AI_MODEL]