Maintained field reference
Practical resources for accountable AI use
Guides, learning notes and research briefings selected for people making real decisions about AI at work. Every entry names its publisher and editorial review state.
AI Disclosure and Use Guide
A free, practical guide to disclosure, permission, verification and staff communication.
Free, ungated and CC BY 4.0.
- Publisher
- AidGPT
- Editorial review
- Reviewed August 2026
Browse the catalogue
Choose by need, not by novelty
Use one or more filters. The complete catalogue remains available when no filter is set.
14 resources
Guide or tool
AI Disclosure and Use Guide
A free, practical guide to disclosure, permission, verification and staff communication.
10 min starting out AI governanceSafe useVerificationRead guide- Type / status
- Guide or tool AidGPT guidance · CC BY 4.0
- Publisher
- AidGPT AidGPT guidance
- Editorial review
- Reviewed August 2026 Review due February 2027
Guide or tool
MI-GUIDE-002 Verification-First
Commissioning AI-assisted analysis you can defend: a working protocol from two market assessments. Seven phases, nine failure modes with the control that caught each one, and ten requirements for anyone commissioning AI-assisted analysis. Free, CC BY 4.0, six pages.
15 min working practice VerificationAI governanceHumanitarian operationsDownload guide- Type / status
- Guide or tool AidGPT guidance · CC BY 4.0
- Publisher
- AidGPT AidGPT guidance
- Editorial review
- Reviewed August 2026 Review due February 2027
Guide or tool
MI-GUIDE-002 Somalia Case Exhibit
The working draft behind the guidance paper, published in full so the method can be inspected. A ten-section desk-based EMMA produced in a single AI-assisted session, with every evidence tag and flagged gap intact. A March 2026 draft, deliberately not updated, published for methods transparency rather than as current analysis.
45 min working practice VerificationHumanitarian operationsDownload case exhibit- Type / status
- Guide or tool AidGPT case exhibit · March 2026 draft, not updated
- Publisher
- AidGPT AidGPT guidance
- Editorial review
- Reviewed August 2026
Guide or tool
AI Workflow Card Safe Use Guide
A field-ready card for defining a bounded AI task, safe inputs, checks, stop rules and escalation.
15 min working practice Safe useVerificationAI governanceDownload workflow card- Type / status
- Guide or tool AidGPT guidance
- Publisher
- AidGPT AidGPT guidance
- Editorial review
- Reviewed August 2026 Review due February 2027
Learning note
Why AidGPT exists / Shadow AI
A starting point for recognising informal AI use and turning it into visible, supported practice.
8 min starting out AI governanceSafe useRead learning note- Type / status
- Learning note AidGPT learning adaptation of MarketImpact commentary
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Learning note
Use AI's mind, not its memory
Use AI to work from supplied material, identify gaps and improve a draft rather than invent authority.
6 min working practice VerificationSafe useRead learning note- Type / status
- Learning note AidGPT learning adaptation of MarketImpact commentary
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Learning note
What effective humanitarian AI training should achieve
A commissioning lens that prioritises safe behaviour, verification and usable workflows over prompt tricks.
7 min strategic TrainingVerificationSafe useRead learning note- Type / status
- Learning note AidGPT learning adaptation of MarketImpact commentary
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Learning note
SAFE AI and the habits no framework can supply
A practical reminder that governance frameworks need staff who can judge evidence at the keyboard.
6 min working practice AI governanceVerificationTrainingRead learning note- Type / status
- Learning note AidGPT learning adaptation of MarketImpact commentary
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Learning note
The Liar's Dividend and information integrity
A briefing on why synthetic media can undermine trust in both false and authentic evidence.
7 min working practice Information integrityVerificationRead learning note- Type / status
- Learning note AidGPT learning adaptation of MarketImpact commentary
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due November 2026
Learning note
Local AI and the disappearing infrastructure excuse
Local models can change data-residency choices, but they do not remove governance, support or verification work.
8 min strategic InfrastructureAI governanceVerificationRead learning note- Type / status
- Learning note AidGPT learning adaptation of MarketImpact commentary
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due November 2026
Research briefing
Infrastructure sovereignty and humanitarian neutrality
A briefing on infrastructure choices as operational, neutrality and accountability decisions.
10 min strategic InfrastructureAI governanceHumanitarian operationsRead research briefing- Type / status
- Research briefing AidGPT research briefing based on a MarketImpact report
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Research briefing
Responsible AI adoption in complex environments / NRC Ukraine
A briefing on responsible adoption in a complex response, including capability, infrastructure and accountability.
12 min strategic AI governanceHumanitarian operationsTrainingRead research briefing- Type / status
- Research briefing AidGPT research briefing based on a MarketImpact report
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Research briefing
Generative AI for humanitarian MPCA
A briefing on where generative AI may assist cash programming and where governance and human review remain essential.
10 min strategic Humanitarian operationsAI governanceData protectionRead research briefing- Type / status
- Research briefing AidGPT research briefing based on a MarketImpact report
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Research briefing
Consent, identity matching and provider trust
A briefing on consent, identity matching and the trust conditions for sharing data with AI providers.
10 min strategic Data protectionHumanitarian operationsAI governanceRead research briefing- Type / status
- Research briefing AidGPT research briefing based on MarketImpact analysis
- Publisher
- MarketImpact Adapted for AidGPT practice
- Editorial review
- Reviewed August 2026 Review due February 2027
Editorial relationship
From MarketImpact
MarketImpact remains the canonical publisher of its research, analysis and commentary. AidGPT selects a small number of sources and turns them into clearly labelled learning notes or research briefings for responsible practice.
Occasional updates
AidGPT Updates
A restrained email note on practical AI adoption, training, governance and newly reviewed resources.