(cases, projects, authentic tasks),oracy(short, targeted interviews),
andversioning(successive drafts with revision notes). This reduces the weight of the “perfect text” and increases that of argumentation and metacognition.artificial intelligence high schoolWhat to document? A simple, repeatable, verifiable policy: (1) whether AI was used, (2) for which phases (ideation, revision, translation, summarization), (3) which prompts or general instructions, (4) which human checks were carried out (source verification, corrections, integrations). The declaration must not be punitive: it is part of digital citizenship education.
To prevent and manage misuse, avoid mere “plagiarism hunting” and build robust assessments: questions that require links to lessons covered, references to lab experiences, critical discussion of typical AI errors, and micro-interviews for confirmation. If inconsistencies emerge, the procedure should be clear and proportionate: request for explanation, supplementary submission, possible guided reworking. Consistency and transparency are part of
.conscious usePrivacy, data, and human oversight: how to be compliant (school and university)transparencyThe use of AI in educational contexts directly impacts personal data, especially when it involves minors. The guiding principle isminimization: include in prompts and materials only what is strictly necessary, avoiding names, identifying details, health or family information. If you need to work on real cases, it is better to anonymize or use fictitious data. Alsoretentionmust be considered: what is saved, for how long, with what access, and for what educational purpose.
In schools and universities, clarify roles: the institution is typically thedata controller, while any service providers may be processors or independent controllers depending on the case. Alignment with the DPO/competent offices is needed, along with a clear notice on purposes and tools used; consent is not always the correct legal basis, but when necessary (especially for external, non-institutional services) it must be handled explicitly and documented.human oversight algorithms schoolHuman oversight concerns not only teaching, but also compliance: establish who reviews outputs, by what criteria, and how errors or inappropriate content are handled. When an output influences decisions (for example evaluative feedback or pathway suggestions), provide for a “second read” and the possibility to challenge it. The keyword is
: being able to reconstruct what was asked, what was produced, and what was decided by the teacher.AI use in class regulationsPractical workflow for teachers: using StudierAI in a compliant, effective, and documentable way
StudierAI
start for freeand define a class policy from the outset. To learn more about the approach and mission, you can find details on theabout uspage.(how it works, limits, bias), and activities in which it is5-step workflow (adaptable to department/class council) for adoptingAI tools for Italian teachers
To develop skills, focus on three axes:Lesson preparation: generate an outline with objectives, prerequisites, and timing; then revise and align it with the PTOF and curriculum planning. Save version and date (log).(source and coherence checks),Differentiation: create variants of the same assignment (basic/intermediate/advanced) and BES/DSA adaptations, checking linguistic accessibility and absence of stereotypes (human check).(what a model “sees,” what it doesn’t) andRubrics and criteria: have it propose a rubric, then customize it with observable indicators (process, sources, accuracy, originality, reflection). Keep the rubric as an attachment to the assignment.(cite, disclose, argue). In secondary school and at university, this translates into assignments that ask not only for a product, but also for a “behind the scenes”: choices, revisions, and motivations.
Inclusion and accessibility: AI can help (language simplification, maps, graded exercises), but it can also amplify inequalities if access is not uniform or if models introduce stereotypes. Build into your design: anti-bias criteria (balanced examples), equivalent alternatives for those who cannot use AI, and accessibility checks (clear language, compatible formats, timing). The goal is that personalization remainsCompliance checklist: before assigning, verify (a) educational purpose, (b) data minimized, (c) planned usage declaration, (d) verification methods (oral/portfolio/versioning), (e) plan for handling errors and challenges., not “automatic.”
- prompt template
- sign up for free
- Use rubrics that assess process (revisions, sources, argumentation) in addition to the final product.
Assessment and academic integrity: what is allowed, what to document, how to verify
In 2026 the question is no longer “how to catch AI,” but how to assess real learning in a world where AI exists. The most compatible practices are those that make the pathway visible:authentic assessments(cases, projects, authentic tasks),oracy(short, targeted interviews),portfolioandversioning(successive drafts with revision notes). This reduces the weight of the “perfect text” and increases that of argumentation and metacognition.
What to document? A simple, repeatable, verifiable policy: (1) whether AI was used, (2) for which phases (ideation, revision, translation, summarization), (3) which prompts or general instructions, (4) which human checks were carried out (source verification, corrections, integrations). The declaration must not be punitive: it is part of digital citizenship education.
To prevent and manage misuse, avoid mere “plagiarism hunting” and build robust assessments: questions that require links to lessons covered, references to lab experiences, critical discussion of typical AI errors, and micro-interviews for confirmation. If inconsistencies emerge, the procedure should be clear and proportionate: request for explanation, supplementary submission, possible guided reworking. Consistency and transparency are part ofAI use in class regulations.
Privacy, data, and human oversight: how to be compliant (school and university)
The use of AI in educational contexts directly impacts personal data, especially when it involves minors. The guiding principle isminimization: include in prompts and materials only what is strictly necessary, avoiding names, identifying details, health or family information. If you need to work on real cases, it is better to anonymize or use fictitious data. Alsoretentionmust be considered: what is saved, for how long, with what access, and for what educational purpose.
In schools and universities, clarify roles: the institution is typically thedata controller, while any service providers may be processors or independent controllers depending on the case. Alignment with the DPO/competent offices is needed, along with a clear notice on purposes and tools used; consent is not always the correct legal basis, but when necessary (especially for external, non-institutional services) it must be handled explicitly and documented.
Human oversight concerns not only teaching, but also compliance: establish who reviews outputs, by what criteria, and how errors or inappropriate content are handled. When an output influences decisions (for example evaluative feedback or pathway suggestions), provide for a “second read” and the possibility to challenge it. The keyword isauditability: being able to reconstruct what was asked, what was produced, and what was decided by the teacher.
Practical workflow for teachers: using StudierAI in a compliant, effective, and documentable way
A simple way to comply with the principles of the guidelines is to adopt a standard workflow. WithStudierAIyou can structure activities and materials while maintaining a documentable approach, useful both for teaching quality and for internal reporting. If you want to experiment in a controlled context, you canstart for freeand define a class policy from the outset. To learn more about the approach and mission, you can find details on theabout uspage.
5-step workflow (adaptable to department/class council) for adoptingAI tools for Italian teachers:
- Lesson preparation: generate an outline with objectives, prerequisites, and timing; then revise and align it with the PTOF and curriculum planning. Save version and date (log).
- Differentiation: create variants of the same assignment (basic/intermediate/advanced) and BES/DSA adaptations, checking linguistic accessibility and absence of stereotypes (human check).
- Rubrics and criteria: have it propose a rubric, then customize it with observable indicators (process, sources, accuracy, originality, reflection). Keep the rubric as an attachment to the assignment.
- Feedback: use AI to suggest formative comments, but apply a teacher filter (tone, consistency with rubrics, concrete examples). Avoid entering unnecessary personal data.
- Compliance checklist: before assigning, verify (a) educational purpose, (b) data minimized, (c) planned usage declaration, (d) verification methods (oral/portfolio/versioning), (e) plan for handling errors and challenges.
One last suggestion: create 2–3prompt templatedepartment-level templates (for units, rubrics, feedback) and use them consistently. It reduces variability, increases equity, and makes it easier to demonstrate the traceability required by the guidelines. If you have to start from scratch, you can alsosign up for freeand set up an initial set of materials with checks and logs already built into your workflow.
