Privacy note and de-identification

Differentiation Signal is designed as a local static browser workbench for teacher planning.

It helps teachers create a de-identified Signal Brief and copyable prompts for external AI tools. It does not generate AI outputs itself.

The workbench boundary

Differentiation Signal has:

• no accounts;
• no backend;
• no database;
• no uploads;
• no analytics;
• no API keys;
• no direct AI model calls;
• no intentional network requests from the app.

The workbench is designed not to automatically store raw lesson drafts, learner context, Signal Briefs, pasted AI outputs, feedback notes, or exported case records. Browser storage should be limited to non-sensitive preferences or reusable non-student templates.

The information a teacher enters remains their responsibility. Differentiation Signal structures and reflects local user input for planning; it does not decide what should be entered, does not send that input to an AI model, and does not play a role in any later submission to an external AI tool.

What not to enter

• student names;
• student IDs or record numbers;
• exact class lists;
• medical, psychological, or diagnostic claims;
• sensitive family, trauma, health, or legal narratives;
• highly identifying biographical details.

What de-identified context means

De-identified context describes the teaching need without identifying a student.

Use wording such as:

• "a small group needs vocabulary support before writing explanations";
• "some learners can identify the steps but lose the reasoning chain";
• "several students need a scaffold that can be faded during the task";
• "one group needs a higher-challenge pathway that keeps the same learning goal."

Avoid names, labels, and private histories. Focus on observable learner-task mismatches and planning needs.

External AI tools

When a teacher copies a prompt into ChatGPT, Claude, Gemini, or another external AI tool, that tool is outside the Differentiation Signal boundary.

Teachers are responsible for checking the copied prompt, confirming it is de-identified, and following relevant school, system, and jurisdiction policy before using any external AI tool.

Privacy scanner limits

The workbench may warn about possible identifying or sensitive details. This scanner is a safety aid, not a guarantee. It may miss some risks, and it may flag safe curriculum wording by mistake.

Teachers must still review every Signal Brief and prompt before copying it externally.