EU AI Act — Art. 13

AI Instructions for Use

For teachers (deployers) · WNK-IFU-2026-001 · Version 1.0 · Last updated: 23 April 2026

About These Instructions

The EU AI Act (Regulation (EU) 2024/1689) requires providers of AI systems to supply deployers — in this case, teachers — with clear instructions for use. These instructions help you understand what the AI features in Writing, no kidding do, what they cannot do, and what your responsibilities are when using them.

As a teacher, you are the deployer of these AI systems under the EU AI Act. That means you carry responsibilities for using the AI appropriately.

You are always the decision-maker.

AI outputs in this platform are suggestions and drafts. Nothing is communicated to students or recorded as a grade unless you take a deliberate action to do so.

Part 1 — What AI Is Used

Writing, no kidding uses Google Gemini 2.5 Flash, a large language model provided by Google, accessed via Firebase GenKit. The AI is a probabilistic text model. It produces outputs based on statistical patterns in training data. It does not “understand” language in the way a human teacher does.

What this means for you: AI outputs can be wrong, inconsistent, or subtly off-target. Always apply your own professional judgment when reviewing AI output before sharing it with students.

Part 2 — AI Features: What Each One Does

AI-001

Writing Prompt Generator

What it does: Generates writing prompts for your class based on the topic, CEFR level, writing type (e.g. formal letter, discursive essay), and any additional parameters you specify.

Limitations

Prompts are generated fresh each time and may occasionally be repetitive or misaligned with your topic
The AI does not know your specific curriculum, school policy, or exam board requirements
Always review generated prompts before assigning them to students

Your responsibility: Check that the prompt is appropriate for your class level and context.

AI-002

AI Correction Tool

What it does: Reviews a student's writing submission and returns inline corrections with type tags (grammar, vocabulary, structure, register), a feedback summary, strengths and weaknesses, prioritised action steps, and a proposed score (0–100).

Limitations

The proposed score is a starting point, not a final grade — it can be inaccurate, particularly for creative writing or highly contextual tasks
The AI may miss cultural references, intentional stylistic choices, or context you know as the teacher
The AI does not know the student personally, their learning history, or special circumstances

Your responsibility: Review the corrections before applying them — some may be incorrect or unnecessary. Treat the proposed score as a reference point. Never forward AI-generated feedback to a student without first reviewing it.

See also: How AI Correction Feedback Works in the Help Center.

AI-004

Quiz Generator

What it does: Generates quiz questions (multiple choice, fill-in-the-blank, true/false, short answer) based on the topic, CEFR level, and format you specify.

Limitations

Generated questions may occasionally contain factual errors — always review before publishing
Multiple-choice distractors may sometimes be too obvious or too similar
The AI does not know what vocabulary you have taught yet

Your responsibility: Review every question and correct any errors before assigning the quiz to students.

AI-005

Quiz Answer Evaluator

What it does: For short-answer and fill-in-the-blank questions, the AI reviews whether a student's answer should be accepted as correct, considering synonyms, paraphrasing, and spelling variants. Returns a recommended accept/reject flag and a reasoning explanation.

Limitations

The AI may accept answers that are too loose, or reject answers that are equivalent in meaning
The AI cannot account for classroom-specific conventions (e.g. 'we only accept British spelling')

Your responsibility: Review every recommendation. Override the recommendation if your professional judgment differs. No mark is recorded until you confirm it.

AI-006

Writing Scorer — Rubric-Based

Requires your review

What it does: Scores a student's writing against your rubric, returning a level selection, points, confidence, and rationale for each criterion — plus totals, an overall feedback paragraph, weakest/strongest criteria, recurrent errors, and steps to improve.

Rubric and quiz scores are combined into a Suggested Grade shown in the Composite Grade panel. This is clearly labelled as a suggestion — it is not a final grade.

Limitations

The AI selects rubric levels based on text analysis; it may interpret descriptors differently than you intended
A confidence value (0–1) is provided per criterion — lower confidence means the AI was less certain
The AI does not know the student's personal circumstances or learning journey
The AI may be more or less consistent across similar submissions

Your responsibility: Review each criterion score and rationale. Pay particular attention to low-confidence scores. In the Finalize Review panel, the Suggested Grade is read-only — you must enter your own grade before submitting. You can click “Use suggested grade” as a deliberate affirmative act, or type your own value.

AI-007

Student Evaluation Generator

Requires your review

What it does: Generates a narrative evaluation paragraph for a student based on their recent submission performance, common mistake patterns, and optional keywords you provide to focus the evaluation.

Limitations

The narrative is based on aggregated statistics; it does not reflect everything you know about the student
The AI adjusts its tone based on performance scores but may not always match your school's communication culture
The generated paragraph is a draft — it is likely to need personalisation

Your responsibility: Edit the evaluation paragraph before sending it to the student or sharing with parents. Add personal context the AI cannot know (effort, attitude, extenuating circumstances). Do not send AI-generated evaluations without reading and approving them.

AI-008 — AI-012

Gapped Text, Keywords, Definitions, Writing Structure, Rubric Creator

These features generate content for teacher use and do not evaluate students directly.

Limitations

Always review generated content before using it in class
The AI may produce vocabulary or examples at the wrong CEFR level — check against your target level
For the Dictionary feature, check definitions against a published reference for high-stakes vocabulary

Part 3 — What AI Cannot Do

What you might expect	Reality
Consistent grading across all submissions	AI output can vary between calls with identical input
Understanding of your specific curriculum	AI has no knowledge of your school, exam board, or syllabus
Detecting AI-generated writing or plagiarism	The platform does not offer an AI-content or plagiarism detector
Knowing the student personally	AI only sees the text submitted; no learning history or context
Making final academic decisions	That is always your role as a teacher

Part 4 — Your Responsibilities as a Deployer

Under Art. 26 EU AI Act

Use AI within its intended purpose
Use AI features only for the educational functions they are designed for. Do not use AI output for purposes not described in these instructions.
Exercise human oversight
Review AI output before it has any effect on a student. Do not auto-apply AI corrections, scores, or evaluations without your review.
Protect student rights
If a student asks why they received a particular score or feedback, you can explain the AI's role and your own judgment. Students have the right to a human review of any AI-assisted assessment under Art. 86 EU AI Act.
Monitor AI performance
If you notice systematic errors, unexpected outputs, or consistently poor quality in AI features, report this to the platform support team (see Part 5).
Inform students
Students should know when AI has been used to assist in generating feedback or assessing their work.

Part 5 — Reporting Issues

If you encounter AI output that is factually wrong in a way that could affect a student's grade, consistently biased, or producing inappropriate content, please report it:

Email: incidents@writingnokidding.com
Incident form: /report-incident
Academic integrity concerns: contact your institution's academic integrity officer

Serious incidents (significant harm to a student's rights or wellbeing arising from AI use) are handled under EU AI Act Art. 73. We acknowledge reports within 2 working days.

Part 6 — Data Sent to the AI Model

AI Correction Tool (AI-002)Student writing text only

Quiz Answer Evaluator (AI-005)Question text, student's answer, correct answer text

Writing Scorer (AI-006)Student writing text, rubric structure — no student name

Student Evaluation Generator (AI-007)Aggregated performance statistics and mistake category summaries — no personal identifiers, no raw submission content

No student photographs, biometric data, contact details, or special category personal data are sent to the AI model.

Document reference: WNK-IFU-2026-001 · Version 1.0 · 23 April 2026