Skip to main content

AI policy

Via AI Policy, you manage rules that are enforced server-side for every AI request. The policy is separate from daily quotas, access times, and the Functions tile.

AI policy helps a school use AI in a focused way. An administrator defines the safe baseline for the environment. A teacher can then set additional restrictions within that baseline for their own group or student. For example, AI can be open during a practice lesson, but more limited during a test, a source-based assignment, or for a student who needs more structure.

For teachers: policy per group or student​

Open Groups, choose a group, and expand AI policy on the right. Under Policy for, choose:

  • Entire group to use the same agreements for all students in the group;
  • a student to set additional restrictions for that student only.

You can block AI access, enable Closed AI mode, block internet, images, code execution, or connectors, and set maximums for context messages and interactions. Then choose Save group policy or Save student policy. With Remove custom policy, the group or student falls back to the existing parent policy.

Unconfigured options are inherited. Student policy is added on top of group policy and can only make it stricter. Teachers cannot relax the school’s or environment’s rules. This prevents a local choice from accidentally disabling a school-wide safety measure.

Educational use

Start with the learning goal, not with as many blocks as possible. Block the internet when students must first reason from their own knowledge or provided sources. Limit interactions to practice short, focused feedback rounds. Use Closed AI mode when you want to prevent distraction from tools. Discuss in advance with students which setting applies and why; clear boundaries make responsible AI use teachable.

A student setting is intended for appropriate support, not as automatic grading or punishment. Regularly check whether the exception is still needed and, where appropriate, explain to the student what is changing.

Fixed policy and temporary access​

On the same group page, there is AI access. If the normal group setting blocks AI, you can start or extend a supervised window of 30, 45, or 60 minutes. For such a window, you can set extra restrictions, such as no internet, images, code execution, or connectors, and limits for context and interactions. Authorized users can also schedule a time window.

Temporary access only lifts the normal group block. Blocks at the environment or school level, access times, and additional student policy remain in force. This allows a teacher to safely provide room for a lesson without bypassing the school’s protective layer.

Levels and exceptions​

The Environment tab contains the default policy. Under Schools and groups, you only add deviations:

  1. add a school exception;
  2. load groups within that school if needed;
  3. load students within a group if needed;
  4. configure and save only the desired exception.

Schools, groups, and students that are not added inherit the parent policy. When policy is resolved, the order is environment → school → group → student. A stricter block at a higher level cannot be overridden at a lower level.

Availability​

Choose Always on or Off. Access by days and hours is managed separately via Access times. Both checks are applied: if AI policy is off, access times do not grant access.

Limits​

  • Max. previous messages in context limits how much conversation history is sent to the model.
  • Max. context tokens also limits the estimated token volume. The newest messages are retained; you will be notified when older context is omitted.
  • Max. output tokens limits the maximum length of a response.
  • Max. interactions per session stops new turns after the limit is reached.
  • Max. session duration stops new turns after the set time has elapsed.

Leave a field empty to inherit the value from the parent level. Below each field, the admin environment shows the effective value and the source. The lowest positive limit from environment, school, group, and student is binding. Regenerating, document generation, and tool use also count as interactions and use the same limits. The desktop runtime retrieves this policy server-side before each local generation.

Theme alerts​

Set up theme alerts in the Notifications tab. A theme alert checks prompts and responses for specified keywords. With Keywords and contextual AI analysis, the analysis can also recognize synonyms and provide contextual relevance, reason, and a safe summary. Choose a theme, detection method, keywords, severity, and action:

  • Flag creates an alert but lets the request continue.
  • Block stops the request.
  • Escalate creates an alert and sends it via the configured in-app and/or email channels to the configured recipients.

Use Add preset to quickly add an editable base for self-harm, violence, bullying, or substance use. Always check the keywords, severity, action, summary instruction, and recipients for your own educational context. A preset only becomes active after you save the policy.

A free model output never blocks on its own: only a deterministic keyword match can enforce the Block action. Contextual analysis can, however, create an additional signal or explain a possible false positive. For each theme, provide a neutral summary instruction, set a retention period, and for escalation configure only recipients who may act within the allowed school scope.

When a response with blocking policy is returned, streaming is temporarily delayed. The full response is checked first and only then stored and made visible. A blocked response is not stored in the conversation or version history.

See Following up on theme alerts for filters, status transitions, ownership, and privacy.

Closed educational context​

Closed educational context is a strict baseline profile for correctional or otherwise highly controlled educational settings. By default, it blocks voice chat, internet and image search requests, image and document generation, code generation and execution, MCP/connectors, and help with hacking or bypassing.

Lower levels cannot lift these blocks. You can also enter an additional policy instruction; it is added server-side alongside the regular system prompt. Availability, access times, limits, and alerts are configured separately.

In AI-School, closed mode also removes the following items from the interface and blocks them on the server:

  • voice chat and transcription;
  • file uploads and File Management;
  • assistants;
  • support requests.

Text chats and their history remain available. As soon as a chat message in closed mode is completed, a student can no longer overwrite the prompt, the answer, or the saved versions. Regenerate therefore creates a new, linked chat message with a version audit label; the original message remains unchanged. If closed mode is later turned off, messages previously created in closed mode remain immutable.

After turning off closed mode, the normal options become visible again. The server rechecks the policy for, among other things, real-time connections, transcription, uploads, and support requests; hiding a menu option is therefore never the only protection.

Checking​

After making a change, start a new chat with an account at the relevant level. Check both the allowed behavior and the expected block, and for theme alerts also review the alert overview.

Assistants for students​

Determine whether students may create their own assistants. Create own assistants · Create assistants with AI

Inheritance follows the policy of the environment, school, and group. Without explicit policy, the permissions apply. A block at a higher level remains in force. Creating with AI also requires permission to create your own assistants.

On this page