Feature · Document · Written by Yobel Tzegai · Last checked 22 October 2026
In Complipath (complipath.io), four documents are drafted from your register by rules, and no model writes them: an AI use policy, Article 50 notices, a human oversight procedure and instructions for users. Every sentence names its source, a gap stays in brackets until you answer, each change makes a new version, and a person approves one.
Drafts an AI use policy for the company and, for each AI system, Article 50 notices, a human oversight procedure and instructions for users, by rules from the register and your answers, with the source named on every sentence. Each change is a numbered version, a person who can edit the workspace approves one, and each version downloads as PDF or Word.
| Document | For | What it asks you |
|---|---|---|
| AI use policy | The company | Who people ask about AI at work, and what data must never go into an AI tool |
| Transparency notices (Article 50) | One AI system | Whether it talks or writes to people, generates content, or recognises emotions or sorts people by biometric data |
| Human oversight procedure | One AI system | Who oversees it, and how a person stops it or overrides what it produces |
| Instructions for users | One AI system | What it must never be used for, and for a high-risk system where the provider's instructions are |
The approved AI use policy can become the policy that everyone on your AI literacy list reads and accepts, with one press. See the literacy paths.
| Paragraph | Who owes it | What it asks |
|---|---|---|
| Article 50(1) | Providers | Design AI systems intended to interact directly with people so that the people are informed that they are interacting with an AI system, unless that is obvious to a reasonably well-informed, observant and circumspect person |
| Article 50(2) | Providers | Mark the outputs of AI systems that generate synthetic audio, image, video or text in a machine-readable format, detectable as artificially generated or manipulated |
| Article 50(3) | Deployers | Inform the people exposed to an emotion recognition system or a biometric categorisation system of its operation, and process personal data under the data protection rules |
| Article 50(4) | Deployers | Disclose that content is artificially generated or manipulated: a deep fake (image, audio or video) and text published with the purpose of informing the public on matters of public interest |
| Article 50(5) | Whoever gives the information in paragraphs 1 to 4 | Give it in a clear and distinguishable manner at the latest at the time of the first interaction or exposure, in line with the applicable accessibility requirements |
Article 14(4) asks that a high-risk system be provided to the deployer so that the people given its oversight are enabled, as appropriate and proportionate, to do what the table lists.
| Provision | To be enabled to | In the draft |
|---|---|---|
| Article 14(4), point (a) | Properly understand the relevant capacities and limitations of the system and duly monitor its operation, including to detect and address anomalies, dysfunctions and unexpected performance | Drafted from the law |
| Article 14(4), point (b) | Remain aware of the possible tendency of automatically relying or over-relying on its output (automation bias) | Drafted from the law |
| Article 14(4), point (c) | Correctly interpret its output, taking into account, for example, the interpretation tools and methods available | Drafted from the law |
| Article 14(4), point (d) | Decide, in any particular situation, not to use the system or otherwise to disregard, override or reverse its output | Drafted from the law |
| Article 14(4), point (e) | Intervene in its operation or interrupt it through a “stop” button or a similar procedure that lets the system come to a halt in a safe state | Not drafted from the law: the stop line comes from your answer or the system's profile |
Every plan, including the free first system. The plan never limits the drafts. Saving and approving a version needs someone who can edit the workspace.
€499 a month
Prices exclude VAT.
No. Rules write each draft from your register and your answers, and every sentence names its source: the register, the law with its article, your answer, Complipath's standard wording or Complipath's records. A person at your company approves a version. The drafts are not legal advice.
Not for every one. Article 50 covers systems that interact directly with people, generate synthetic content, or recognise emotions or categorise people by biometric data. The draft asks per system and, if none applies, says Article 50 asks for no notice for it. Answer again if the use changes.
A saved version stays as it was. The page tells you the register has changed since that version and shows the new text; saving it makes the next numbered version, which needs its own approval. The latest eight versions stay listed, each with a PDF and a Word file.
Every plan, the free one included: the plan never limits the drafts. Free holds one AI system, with no end date. Starter is €499 a month for up to 20 AI systems. Prices exclude VAT. Saving and approving needs someone who can edit the workspace. Growth and Scale are not available yet, and nothing here waits for them.