UK MOD JSP 936 (dependable AI in defence)

What it is

The UK Ministry of Defence published JSP 936: Dependable Artificial Intelligence (AI) in defence (part 1: directive) on 13 November 2024. GOV.UK calls Part 1 the principal policy framework governing the safe and responsible adoption of AI in MOD.

The published PDF is JSP 936 V1.1. The preface says the JSP mandates the application of ambitious, safe and responsible practices relating to all Defence projects that include Artificial Intelligence (AI). It is designed to be used by MOD staff responsible for developing and/or using systems which include AI, regardless of the application domain, and will be reviewed at least every two years. The owner is the MOD Director General Delivery & Strategy; it is managed by the Defence AI and Autonomy Unit (DAU) and the Defence AI Centre (DAIC).

GOV.UK says the directive provides clear direction on how to implement the MOD’s AI ethical principles and includes directives on governance, development and assurance throughout the AI lifecycle which encompasses quality, safety and security considerations.

The five ethical principles it implements

JSP 936 builds on our AI Ethical Principles, set out in the Ambitious, Safe and Responsible Policy Paper (ASR). The JSP copies those five principles verbatim from the ASR policy:

1. Human-centricity — The impact of AI-enabled systems on humans must be assessed and considered, for a full range of effects both positive and negative across the entire system lifecycle.
2. Responsibility — Human responsibility for AI-enabled systems must be clearly established, ensuring accountability for their outcomes, with clearly defined means by which human control is exercised throughout their lifecycles.
3. Understanding — AI-enabled systems, and their outputs, must be appropriately understood by relevant individuals, with mechanisms to enable this understanding made an explicit part of system design.
4. Bias and harm mitigation — Those responsible for AI-enabled systems must proactively mitigate the risk of unexpected or unintended biases or harms.
5. Reliability — AI-enabled systems must be demonstrably reliable, robust and secure.

The ASR says these principles do not affect or supersede existing legal obligations.

Must, should, and who it applies to

Where the JSP says must, the action is a compulsory requirement. Where it says should, the action is not a compulsory requirement but is recommended good practice.

The requirements set out in this JSP must be implemented from the outset of all Defence Capability development where MOD requirements lead to the use of AI. For legacy AI, alignment with the requirements of the JSP should be achieved in so far as is possible. GOV.UK also tells industry partners they can use Part 1 to align projects with defence expectations.

Part 2 is guidance. MOD’s AI Practitioners’ Hub provides guidance and resources to help you comply with the policy and requirements outlined in JSP 936 Part 1.

What it is not

JSP 936 does not provide a list of AI-enabled systems or capabilities which are compliant or not compliant. It is UK MOD policy, not a US CMMC level and not the DoD Responsible AI toolkit. The directory lists the AI Practitioner’s Handbook as the official Part 2-style resource. For what Part 1 says to a supplier, see What JSP 936 Part 1 asks of industry partners and suppliers. For a sourced comparison, see UK MOD JSP 936 vs DoD Responsible AI.

Primary sources

Read the GOV.UK publication page and the Part 1 PDF. The Defence Artificial Intelligence Strategy (15 June 2022) is the strategy the JSP follows.