Data, Security & Governance
AI permissions, plug-ins, and integrations
Review what an AI tool can access, store, share, or change before enabling it.
5 min readGovernance
Workplace example
Email and file access
A third-party AI plug-in asks for email and file access. Check whether it is approved, what it can read, and what review is required.
What this means
- •An AI plug-in or integration may ask for access to email, files, calendars, customer records, or internal systems.
- •The risk is not only the prompt. It is also what the connected tool can read, store, share, or act on.
- •Permission requests should be necessary for the business use case and approved through the right route.
Why it matters
- •Broad permissions can expose more information than the task requires.
- •Summarising a document still requires the same access controls as viewing it.
- •Unfamiliar third-party tools can create data, security, privacy, and compliance risk.
Common mistakes
- •Enabling a tool because the brand or interface looks professional.
- •Trying it with a small sample of work data before approval.
- •Assuming access controls are only a technical team concern.
What good judgement looks like
- •Check approval status before enabling an integration.
- •Review requested permissions against the actual use case.
- •Keep normal access permissions in place for internal AI assistants.
Try this at work
- •Pick one tool or plug-in you have seen.
- •Write what it asks to access.
- •Decide what approval or review would be required before use.
How this helps your reassessment
- •You know what to check before enabling an AI plug-in.
- •You understand that access permissions still matter when AI summarises content.
- •You avoid unfamiliar third-party services until approval and risk are clear.
Related guides
Public, internal, and organisation-run AI tools
Know why different AI tools need different data, access, and approval checks.
Sensitive data and approved tools
Know what information must not be entered into unapproved AI tools and why approval matters.
Responsible AI workflow design
Design AI-assisted workflows with clear use cases, boundaries, review points, and ownership.