Studying Cyber Resilience beyond Technical Controls
Frame cyber resilience as an institutional capability involving people, governance and recovery.
GRS Editorial · GCAIRead guideGRS editorial · Digital research
Responsible AI Use in Institutional Research
Use AI to support research workflows while keeping evidence, confidentiality and accountability with people.
Digital research
A practical starting point.
Use AI to support research workflows while keeping evidence, confidentiality and accountability with people. This guide offers a starting method for teams developing research and institutional decisions. Adapt it to the question, context and relevant scientific, ethics and data requirements.
Step 01
Classify the task and data
Literature discovery, transcription, coding assistance and drafting carry different risks. Before using a tool, identify whether inputs contain personal, confidential or unpublished material and what its terms permit. Use an approved environment and remove unnecessary identifiers.
Step 02
Keep source verification human
Generated citations and summaries can be incomplete or fabricated. Open primary sources, verify quotations, check methods and record the exact source that supports each claim. AI-assisted search may widen discovery, but it does not replace a reproducible search strategy or critical appraisal.
Step 03
Document material use
Decide in advance which tasks may involve AI and how that use will be disclosed where relevant. Keep a record of tool, purpose, human review and substantive corrections. Avoid placing unpublished partner information into services whose retention or training settings have not been assessed.
Step 04
Protect the final decision
A named researcher remains accountable for design, interpretation and publication. Review bias, errors, copyright and data protection before release. The useful test is whether the tool improves a specific research step without weakening traceability, consent or quality.
Illustrative application
What this could look like.
A team may use an AI tool to propose search terms for a literature review. Researchers should then build the final search themselves, document databases and dates, and verify every included paper. If interview transcripts contain identifiable information, the team should assess permissions and approved processing arrangements before considering any automated summary service.
Put it into practice
A short working checklist.
- Write the precise decision or research question this work should support.
- Record the sources, methods, stakeholders and permissions needed before proceeding.
- Agree who reviews the output and how limitations will be communicated.
- Set a date to revisit the decision when evidence or conditions change.
GRS Group editorial guide · Published 24 September 2026 · Informational guidance, not an empirical study or formal professional advice.
Related research
Continue with the centre and knowledge model.
Global Cybersecurity & AI Research Center Research & Knowledge All guidesFurther reading