USE AI SAFELY
Use AI with confidence.
Know what to share, what to check and what to keep under your control.
AIstratica Editorial · Updated 26 October 2026
AI can help you learn, create and get work done. These practical checks help you decide how to use a tool for your task. They are guidance, not a safety certification of any product.
1. Share thoughtfully
Start with a sample or anonymised document. Avoid uploading passwords, identity documents, confidential work, student records or identifiable patient information to an unapproved service. Removing a name alone may not make a record anonymous.
Before uploading, check the provider’s data-retention, deletion and model-training settings for your exact plan. Review who can see shared links and connected files. If the policy is unclear, keep sensitive material out.
2. Check what matters
Treat generated answers as a starting point. Open cited sources, verify dates and calculations, and check that a source actually supports the claim. For important decisions, use an appropriate qualified person to review the result.
For creative work, inspect generated images, captions, names and audio before publishing. A convincing result can still contain mistakes.
3. Keep agents under your control
Give an agent only the accounts, files and permissions needed for the job. Start with read-only access where possible. Require your approval before it sends messages, spends money, deletes files or publishes work.
Review its proposed actions and activity history. Treat instructions inside unfamiliar websites and documents as untrusted content. Know how to stop a task and revoke the agent’s access.
4. Understand costs and usage rights
Check monthly versus annual billing, credit limits, overage charges, trial renewal and cancellation steps before subscribing. Keep a record of the plan you selected.
Check the current terms for commercial use, uploaded materials, generated outputs and voice or likeness permissions. Do not assume a paid plan grants rights to every input or output.
5. Match the safeguards to the setting
Education
Use AI to explain ideas, practise skills and improve drafts while keeping the learner’s own reasoning central. Follow the school’s rules on permitted assistance and disclosure. Check age eligibility and consent arrangements; protect student information. Teachers should review generated lessons and assessments before use.
UNESCO guidance on generative AI in education and research supports privacy protection and human-centred, age-appropriate use.
Healthcare and medical research
Use research tools to help locate evidence and prepare questions. Read the underlying studies and check their relevance. AI output does not replace professional diagnosis or treatment; do not change medication or delay care based on a chatbot answer.
Clinical tools need assessment for their intended use, local requirements and institutional approval. Keep identifiable patient data within approved systems, and have clinicians review generated notes and recommendations. A clinical marketing claim alone is not proof of validation.
WHO’s guidance on safe and ethical AI for health explains why plausible answers can still be inaccurate and require careful oversight.
A quick check before you begin
- Can I complete this task without sharing sensitive data?
- Which outputs will I verify, and against what source?
- Have I limited access and kept approval for consequential actions?
- Do I understand the price, cancellation terms and usage rights?
- Does this setting require a teacher, clinician or other qualified reviewer?