AI Principles - Papyria

1. Purpose

Papyria uses artificial intelligence (AI) to support copy-editing, proofreading, translation, and typesetting. The goal is more efficient production and consistent quality — without compromising professional judgement, confidentiality, or the customer's guidelines.

AI should be a tool, not a replacement for expertise.

2. Core Principles

2.1 Gentleness

We preserve meaning, message, tone, and style.
We avoid unnecessary rewrites and 'over-improvement'.
We follow the customer's style guide, terminology, and editorial guidelines.

2.2 Verifiability

AI contributions must be documentable and reviewable after the fact.
We ensure traceability where appropriate, for example with tracked changes, comments, or versioning.
We can disclose which tools were used, when relevant.

2.3 Quality Assurance

AI is neither the first nor the last step in the process.
The result must be reviewed by a person with relevant expertise.
We pay special attention to typical AI errors, such as inconsistencies and 'hallucinations'.

2.4 Transparency

We are clear with customers about when and how AI is used, when this is relevant.
If the customer wants 'no AI' in a project, we accommodate that.
If AI is to be used to generate new content (not just processing), we explicitly disclose this.

3. Targeted Use

We use AI when it provides clear added value, without compromising safety and quality. This means we:

use AI purposefully, and no more than necessary to complete the task
choose other methods when they provide better precision or control

3.1 What We Do

We use AI primarily for safe changes, suggestions, and repetitive tasks — not as the final word.
We share only what is necessary with AI tools (data minimisation).
We stop and clarify internally if we are uncertain about confidentiality, rights, or customer requirements.

3.2 What We Don't Do

We do not input customer material into open/unapproved AI services where we have no control over how data is processed.
We do not present AI-generated sources or claims as if they were verified.
We do not use AI to imitate named authors, translators, or other contributors with the purpose of replacing their work.

4. Responsibility and Updates

Everyone working on behalf of Papyria (employees and freelancers) must follow these principles. The policy is updated as needed.

5. Contact

Questions: support@papyria.com