AI statement

What the AI does

Kuraplan uses large language models to help teachers plan lessons, units and weekly programmes, and to draft classroom resources such as worksheets, slides, quizzes and activities. The teacher describes what they need; the assistant drafts it against the teacher's chosen curriculum; the teacher edits and decides what to use.

Outputs are text, worksheets, slides, quizzes, printable activities and illustrative images. Teachers can paste in their own material to be adapted.

What the AI is not used for

• Processing personal information about students or anyone else. Prompts are meant to describe topics, year levels and learning goals, not people.
• Profiling, monitoring or assessing students.
• Making decisions about any person.
• Biometrics, emotion detection, health data, or reading files from connected drives.
• Training or fine-tuning models. Your prompts and outputs are never used to train any model, ours or a provider's.

Who can use it

Signed-in teachers and adults for planning and resource creation. Students, without accounts, can use two features inside a teacher-shared activity: a helper chat that guides without giving answers, and hints. Those messages, the question and the student's answer go to the model; the student's name does not. The AI also marks answers for the teacher. Real-time detection and alerting for harmful content in student messages, and a crisis-response protocol, are not yet in place and are being built.

Models and providers

Requests are routed through Vercel AI Gateway to Anthropic, OpenAI and Google models hosted in the United States, under each provider's business terms with no training on customer data. We publish the current list on the Subprocessors page and give at least 30 days' notice before changing the country where models run.

Limits

Generated content can be wrong, out of date, or biased, and can misread a curriculum statement. Every output is labelled as AI-generated. Treat it as a first draft: check facts, check curriculum alignment, and check suitability for your students before use.

Keeping personal information out

Please do not put student names, family details or other personal information into prompts or uploads. We are adding automatic filtering that removes names and personal details before a request reaches a model, along with a reminder under the prompt box.

Logging and retention

Each request and output is logged with a timestamp and account identifier for 90 days so we can investigate faults and abuse, then deleted. You can delete your conversation history at any time.

Your controls

• Delete conversations from within the product.
• Turn personalisation on or off in settings.
• Opt out of product news at any time.
• Ask for your account to be deleted.

Reporting a problem with an output

If the assistant produces something inaccurate, biased, offensive or otherwise inappropriate, use the report control on the output or email security@kuraplan.com with the resource link. We review reports within two business days, remove content that should not have been generated, and adjust filters and prompts to stop it happening again.

Governance and testing

The founder is accountable for AI safety and reviews AI processing purposes against this statement each quarter. We are establishing regular jailbreak and prompt-injection testing, privacy tests that confirm personal details do not reach providers, and an evaluation set for bias and content integrity across year levels and cultures. Reports from each round are kept for assessment.