Thalius Thalius

Legal & trust

AI Act disclosure

Effective 2026-10-19 · version 2.0

You are interacting with an AI system

Thalius Hippocampus is an AI system within the meaning of Regulation (EU) 2024/1689 (the AI Act). Answers in the product are generated by a large language model from passages retrieved from the documents you or your organisation uploaded or imported. Every answer is marked as AI-generated and cites the passages it was generated from, and a question the documents cannot answer is answered as not found in your documents, or clearly labelled as general advice.

Our role

Thalius AI AB (organisation number 559528-5635, VAT SE559528563501, Slipgatan 5, 117 39 Stockholm, Sweden) is the provider of the Hippocampus AI system. We do not develop general-purpose AI models; the models that generate answers are supplied by the sub-processors named on the sub-processors page and are selected per workspace. Workspaces that publish a public chat or connect apps to the Service are deployers for that use.

Transparency obligations (Article 50)

The transparency obligations of Article 50 apply from 2 August 2026.

  • The product marks every answer as AI-generated and shows the sources it was generated from.
  • Public chats show visitors that they are talking to an AI assistant, who publishes the chat, and that spoken answers are synthetic voice.
  • The Hippo Guide is an AI assistant answering from our documentation. Its answers describe the product, are not advice, and are not part of the contract. Using the Guide for anything other than questions about Hippocampus and the documentation Thalius provides is not an agreed use under the terms of service.
  • Machine-readable marking of AI-generated text within closed enterprise deployments is under legal assessment; this page will be updated when the position is settled.

What the system is not for

Hippocampus retrieves and cites what your documents say. It does not make decisions about people, and it is not intended for use as the sole basis for decisions with legal or similarly significant effects on individuals. It is not designed for any high-risk use listed in Annex III of the AI Act. A customer who deploys it for such a use takes on the deployer obligations for that use, and should tell us first.

Human oversight

Every answer shows the passages it came from, so a person can check it against the source in one click. Workspace owners control which documents the system can read and which model generates answers. People can correct the knowledge base directly. The terms of service require users to review output before relying on it.

Your documents and the models

We do not train models on customer documents. Model providers receive only the fragments needed for a specific answer, transiently. Each provider is named on the sub-processors page.

Questions

Write to privacy@thalius.ai, or contact us through the Corporate route.