Research AI Landscape

DR-01-governance-gdpr

3 min read

DR-01 — Governance, GDPR & Accessibility

Purpose: Resolve data-training-policy, gdpr-compliant, data-residency, and accessibility-statement unknowns for AI tools under evaluation at the University of Sheffield.

Use with: Claude Research mode (or equivalent deep research tool)

Prompt

You are helping a UK research university (University of Sheffield) evaluate AI tools for institutional procurement. For each tool listed below, research and return the following information.

For each tool, find:

  1. Data training policy — Does the tool use user-submitted content to train or improve its AI models? Look for: privacy policy sections on model training, “opt-out” provisions, enterprise/institutional agreements that disable training, and any published DPA (Data Processing Agreement) terms. Classify as one of:

    • no-training — vendor explicitly commits that user data is never used for model training
    • private-training — user data may be used to improve models but only for that organisation (federated/private)
    • public-training — user data may be used to train shared/public models
    • unknown — policy is absent or ambiguous

  2. GDPR compliance — Is the tool GDPR-compliant for European/UK users? Look for: EU/UK GDPR compliance statements, SCCs (Standard Contractual Clauses), UK IDTA, ISO 27001 certification, SOC 2 reports, and whether a DPA is available. Classify as yes, no, or unclear.

  3. Data residency — Where is research data processed and stored? Identify the primary region(s): EU, UK, US, unknown, or multiple.

  4. Accessibility statement — Does the vendor publish an accessibility statement or WCAG conformance report? Classify as yes, no, or unknown. If yes, note the claimed conformance level (e.g. WCAG 2.1 AA).

Tools to research:

  • Grammarly (grammarly.com) — academic writing assistant
  • Writefull (writefull.com) — academic writing tool for researchers
  • Elicit (elicit.com) — AI research assistant for literature review
  • Consensus (consensus.app) — AI-powered academic search
  • Scite (scite.ai) — smart citation analysis platform
  • Dimensions AI (dimensions.ai) — research intelligence platform
  • Perplexity AI (perplexity.ai) — AI-powered academic search
  • Julius AI (julius.ai) — AI data analysis tool
  • SciSpace / Typeset (scispace.com) — AI research platform
  • Zendy (zendy.io) — AI-powered academic library
  • Covidence (covidence.org) — systematic review platform
  • Rayyan (rayyan.ai) — systematic review and screening tool
  • ResearchRabbit (researchrabbitapp.com) — literature mapping tool
  • NVivo (lumivero.com/nvivo) — qualitative data analysis
  • MAXQDA (maxqda.com) — mixed-methods research software
  • ArcGIS Pro with GeoAI (esri.com) — GIS with AI features
  • Google Co-Scientist (Google DeepMind) — AI for scientific hypothesis generation
  • Kosmos (kosmos.io or similar) — research AI platform
  • PaperQA2 (github.com/Future-House/paper-qa) — open-source LLM for scientific PDFs
  • Grantable (grantable.co) — AI grant writing assistant
  • Granted AI — AI grant writing assistant

Output format:

Return a markdown table with columns: Tool | data-training-policy | gdpr-compliant | data-residency | accessibility-statement | Notes

Then provide a short paragraph for any tool where the classification required significant interpretation or where the policy is unclear and a manual review is recommended.

Priority note for the University of Sheffield: The university processes special category research data (health records, sensitive social science data). Any tool classified as public-training or gdpr-compliant: no must be flagged prominently as unsuitable for sensitive research data use cases.

Graph View