Responsible AI

1. Our AI Principles

• Human-in-the-loop by design: every AI suggestion is presented for you to accept, edit, or reject — never applied automatically to your submitted work.
• Built for academic context: prompts and tools are shaped around real assignment, research, and study workflows, not generic chat.
• Transparency: we tell you when a feature is AI-generated, and we explain its limitations in-product.
• Data minimisation: only the content needed to complete your request is sent to an AI model provider.

2. How AI Is Used in the Workspace

AI models power features such as outline generation, grammar and rewrite suggestions, flashcard and quiz creation, research topic and gap suggestions, and the AI Tutor's explanations. Each of these is a suggestion for you to review, not a finished, submission-ready output.

3. Automated Document Processing

Document tools such as Document Studio combine deterministic formatting rules with AI used mainly to interpret ambiguous content — for example, whether a bold paragraph is really a heading, or how a reference should be parsed. Straightforward, high-confidence corrections are applied automatically; ambiguous or low-confidence cases are flagged for you to review rather than changed silently.

In other words, the tool corrects what it can confidently correct and identifies the rest. It does not guarantee that every issue has been found or fixed, and it is not a guarantee of compliance with a specific institution's requirements. Always review the corrected document — and the items flagged for review — before you submit it.

4. Model Providers

We use established third-party AI model providers under contract, selected and configured for the task at hand. Providers are contractually prohibited from using your content to train their general-purpose models.

5. Known Limitations

• AI output can be inaccurate, outdated, or overly confident — always verify facts and citations independently.
• AI models can reflect biases present in their training data; report anything that seems unfair or inappropriate.
• AI-generated summaries or explanations may oversimplify nuanced academic material.

6. Feedback and Reporting Issues

If an AI feature produces something incorrect, biased, or inappropriate, please report it to support@sizapoint.com or through the Contact page — this directly informs how we tune and improve our prompts and safeguards.