JSP 936 V1.1 Dependable Artificial Intelligence (AI) in Defence (part 1: directive)

• Authority: Government of the United Kingdom
• Jurisdiction: Other countries
• Collections: Miscellaneous documents
• Most recent activity: Enacted
• Most recent activity date: 2024-11-01

Risk subdomain coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the risk subdomain are explicitly described.

• Unfair discrimination and misrepresentation
• AI system security vulnerabilities
• Overreliance and unsafe use
• Loss of human agency and autonomy
• Governance failure
• AI pursuing its own goals
• Lack of capability or robustness
• Lack of transparency or interpretability

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• Unequal performance across groups
• Compromise of privacy
• False or misleading information
• Disinformation, surveillance, and influence at scale
• Cyberattacks, weapon development or use
• Increased inequality and decline in employment
• Competitive dynamics
• Environmental harm
• AI possessing dangerous capabilities
• Multi-agent risks

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has good coverage of approximately 12-14 subdomains, with strong focus on AI system security (2.2), malicious actors (4.1, 4.2), governance failure (6.5), AI safety failures (7.1, 7.2, 7.3, 7.4), and multi-agent risks (7.6). Coverage is concentrated in security, misuse prevention, AI safety domains, and governance. There is minimal coverage of discrimination/toxicity and misinformation domains.

Sector coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the sector are explicitly described.

• Scientific Research & Development Services
• National Security

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the sector are described.

• Information
• Professional & Technical Services
• Management, Administrative & Support Services
• Educational Services

All other sectors had no coverage in this document.

Sectors Summary

This document primarily governs AI use within the National Security sector (military defence operations and capabilities). It also has secondary coverage of several support sectors including Professional and Technical Services, Scientific Research and Development Services, Information (for digital systems and data processing), and Management/Administrative Services that support defence operations.

Legislative Status

Hard Law

Legislative Status Summary

This is a binding internal military directive (JSP - Joint Service Publication) with mandatory requirements, enforcement mechanisms, and clear accountability structures. The document uses mandatory language throughout ('must', 'shall') and establishes formal governance, approval, and compliance mechanisms including ministerial oversight and parliamentary accountability.

Actors

Proposer

• Entity Types: AI_Governance_Actor
• Summary: The document is a Joint Service Publication (JSP) issued by the UK Ministry of Defence. The Defence AI Unit (DAU) is identified as responsible for setting AI policy in relation to ethics, and the Defence AI Centre (DAIC) provides technical guidance and direction.

Target

• Entity Types: AI_Developer, AI_Deployer, AI_User
• Summary: The JSP explicitly applies to all MOD staff across all phases of the AI lifecycle and identifies specific stakeholder groups. It also applies to suppliers delivering AI products to MOD, making them subject to the requirements.

Enforcer

• Entity Types: AI_Governance_Actor
• Summary: The document establishes a clear governance hierarchy with enforcement responsibilities. TLB Holders and Chief Executives are responsible for issuing direction and ensuring compliance. RAISOs ensure appropriate assurance is in place. High-risk projects must be referred to JROC/IAC or Ministers for approval.

Monitor

• Entity Types: AI_Governance_Actor
• Summary: The document establishes monitoring through annual assurance statements, ongoing risk management, and dedicated monitoring roles. TLB Executive Boards must provide annual Statements of AI Ethical Assurance. RAISOs oversee ethical development and use. Independent Ethics Assurance mechanisms are required for significant projects.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Plan and Design
• Collect and Process Data
• Build and Use Model
• Verify and Validate
• Deploy
• Operate and Monitor

AI Lifecycle Summary

The document comprehensively covers all AI lifecycle stages from planning through to operation and monitoring. It provides detailed requirements for each stage including planning, requirements, architecture, algorithm design and implementation, data collection and preparation, model development, verification and validation, integration and deployment, and ongoing monitoring. The document emphasizes through-life management and continuous assurance.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• AI Systems
• Generative AI
• Open Weight / Open Source
• Task-specific AI

Technical Scope Summary

The document explicitly covers AI models, AI systems, and various AI types. It provides a general characterization of AI as 'Machines that perform tasks normally requiring human intelligence, especially when the machines learn from data how to do those tasks.' The document addresses both general-purpose and task-specific applications, mentions generative AI (including Large Language Models), and discusses machine learning extensively. It does not use compute thresholds or explicitly discuss open-weight models, though it addresses third-party software and open source tools.