Using AI can introduce data security and privacy risks. These steps can reduce risk, but no tool or checklist can eliminate it.
Know what you are sharing
Identify personal, confidential and commercially sensitive information. Share only what is necessary and authorised. Removing names does not always make information anonymous if a person can still be identified.
Choose and approve tools
Review the provider’s terms, security controls, retention, access, support, incident process and how information is used. Consider the type of task and data rather than assuming paid or free tools are automatically safe or unsafe.
Limit access and monitor
Give people and AI tools access only to the information and actions they need. Keep appropriate records and alerts, and agree who can investigate or disable a tool. Protect any logs that contain sensitive information.
Review and get specialist advice
Test outputs, train staff and plan how to respond to errors or incidents. Data protection obligations depend on the use and information involved; involve your privacy and security advisers where needed.
Ask before connecting information
Identify the source, who owns it and the minimum information the task needs. Check who can access the tool, where information is stored and how it can be removed. Do not assume a document is safe to share simply because a name has been removed.
Test permissions with people who should and should not have access, and agree who can disable the connection or investigate a problem. Review supplier terms and internal guidance as the tool and its use change.
The ICO’s AI and data protection guidance is a starting point for personal-information questions.