AISOF

Product track

AISOF-3380

AI in Defence and Security Systems

Responsible control of military and national security AI.

Version 1.0. Published .

Bright modern defence operations room with human operators monitoring autonomous systems status walls, AISOF-3380 defence systems AI framework

AISOF-3380

Why AISOF-3380 matters

Human control and governability for defence and security AI

AISOF-3380 is the AISOF® product framework for AI used in defence and security systems: weapons-adjacent systems, autonomy, ISR, surveillance, EW, logistics, and decision support. Focused on human control and governability, not capability scoring.

Overview

Defence and national security AI sits under classification, programme security, and political scrutiny. Public AI ethics language is not a product-level determination of governability for a named system.

AISOF-3380 does not score targeting accuracy, strike effectiveness, or classified performance. It addresses control and governability under scheme rules and agreed information-handling arrangements.

Outcomes are Certified or not Certified, determined only by Approved AISOF Assessors under those rules and arrangements.

What sets this framework apart

Control over capability claims

Oriented to human judgement, abort, safe failure, and governability, not a public score for classified performance.

Named system scope

Product-level: the system under assessment, not a free pass for an entire force structure or vendor catalogue.

Binary certification

Certified or not Certified under scheme rules and handling arrangements appropriate to the domain.

Assessor determination

Approved AISOF Assessors determine outcomes under the scheme.

Not generic cybersecurity

Not a substitute for ordinary IT security certification or classified capability evaluation.

Handling-aware scheme

Built for a domain where assessment may require controlled information practices. Full materials are not a public download.

Who seeks AISOF-3380

  • Makers of AI systems used in military or national security contexts
  • Programmes that must separate governability claims from capability claims
  • Teams that cannot rely on consumer AI frameworks or open public checklists
  • Suppliers facing buyers who ask about human control and failure modes

Not for generic enterprise IT. Civilian product sectors use other AISOF product frameworks. Organisational governance uses AISOF-3300.

Public focus themes

The public themes of concern for AISOF-3380.

  • Meaningful human judgement and control
  • Abort, deactivation, and safe failure
  • Data integrity and protection
  • Product security
  • AI supply chain
  • Monitoring and material change
  • Incident and field safety response

Scheme materials. This page describes the purpose, scope and public themes of the framework. The detailed requirements, the assessment method used by Approved AISOF Assessors and the scoring approach are confidential scheme materials, issued under scheme terms with the assessment pack.

Scope: Not generic IT cybersecurity. Does not assess targeting accuracy, strike effectiveness, or classified capability details.

Related frameworks

Assessment packs for AISOF-3380

Request the pack for this framework under scheme terms.