Privacy, confidentiality, copyright, policies and model limitations

Protecting personal and confidential information

Anything you type or upload into an AI service is sent to its provider. Depending on the service and your settings, it may be stored, reviewed by people for safety or quality, or used to improve models. Policies differ between free consumer tools and workplace or education versions, and they change — read the current terms of the tool you use and follow your organisation's rules.

Do not paste (unless you are authorised and the tool is approved for it):

• passwords, access keys, card or bank details;
• other people's personal data — names with contact details, health information, ID numbers, private messages;
• confidential business information — unreleased plans, contracts, client data, proprietary code;
• anything you have promised to keep confidential.

Anonymise instead: replace names with roles ("Customer A", "the manager"), remove addresses, account numbers and dates of birth, generalise specifics ("a mid-sized retailer"), and use fake sample data for structure.

Copyright and attribution

• Your inputs — pasting someone else's copyrighted work into a tool can raise legal and contractual questions; use material you own, have permission for, or that the licence allows.
• Outputs — AI output may resemble existing works. Do not ask for, or publish, close imitations of copyrighted text, images, music or code, and do not present others' work as your own.
• Ownership and protection of AI-generated content varies by country and is still developing. Do not assume you own, or can protect, AI output in every jurisdiction; get qualified advice for commercial use.
• Licences and terms — open-source code and stock assets come with licence conditions; a public web page is not automatically free to reuse.

This lesson is general education, not legal advice.

Policies you must follow

Several sets of rules may apply at once: the AI provider's usage policies, your employer's or school's AI policy, professional codes, platform rules where you publish, and applicable law. Common requirements include disclosure of AI assistance, restrictions on certain data, and bans on certain uses. When rules conflict or are unclear, ask before using AI.

Understanding model limitations

Keep these practical limits in mind when deciding whether to use AI at all:

• No guaranteed truth — outputs must be verified.
• Knowledge cutoff — recent events may be unknown.
• Context limits — very long inputs may be partly ignored.
• Inconsistency — the same prompt can produce different answers.
• Arithmetic and exact counting — unreliable over many steps.
• No accountability — the model cannot take responsibility for decisions; you can.

Stage 3 assessment

Today's completion rule is the Stage 3 assessment, covering Modules 7–10. It unlocks after days 13–19. You can retake it and request a human review.