No product is available yet, so this statement describes how our infrastructure is designed. It is not a declaration of conformity, and we will update it at launch.
The law
The EU Artificial Intelligence Act, Regulation (EU) 2024/1689, as amended by the Digital Omnibus on AI, Regulation (EU) 2026/1744. It applies to us because we intend to offer our services to businesses in the European Union, although we are established in the United States.
Our role
| What we offer | What it is | Our role under the Act |
|---|---|---|
| The pipeline (privacy, memory, context, masking, routing) | Services other companies build their AI products on | A supplier of components. The company that offers an AI system under its own name is its provider |
| The developer tools (command-line tool, SDKs, MCP server) | Ways for those companies and their developers to use the pipeline | Part of the same supply. They are not an AI system offered to the public |
We use general-purpose AI models through their providers' APIs. We do not train, fine-tune or place such models on the market, so we are not a provider of a general-purpose AI model and the Act's obligations for those models do not fall on us.
If we offer an AI system to the public under our own name, we will be its provider, and we will update this statement before it launches.
Risk level
The pipeline is not intended for any of the high-risk uses listed in Annex III, such as employment, education, credit, essential services or law enforcement. Our acceptable use policy forbids using our services to make decisions of that kind unless a person reviews each decision and the laws that govern such decisions are met. A business that builds a high-risk system on the pipeline is its provider and carries the duties the Act sets for it.
What we do
Transparency (Article 50)
- The duties of Article 50 fall on the provider of each AI system: a product built on the pipeline must tell people they are talking to an AI and mark the text it generates.
- The pipeline is designed to support them: every record keeps its source, and every routing decision is recorded, so a product can show where an answer came from.
- We will not present generated content as written by a person.
AI literacy (Article 4)
We take measures to support the AI literacy of our staff and of the teams that build on our infrastructure: plain documentation of what the system does, what it does not do, how masking works and where it can miss.
Prohibited practices (Article 5)
None of the practices the Act prohibits is part of our services, and our acceptable use policy forbids using them for such practices, including the generation of child sexual abuse material and of intimate images of real people without their consent.
Beyond the Act
The design goes further than the Act requires of a supplier of components: personal details are masked before a model sees them, every record keeps its source, the router keeps none of the content, and nothing sent through the pipeline trains a model.
The dates that apply to us
| Date | What applies |
|---|---|
| 2 February 2025 | Prohibited practices and AI literacy |
| 2 August 2025 | Rules for general-purpose AI models (for their providers) |
| 2 August 2026 | Transparency under Article 50, and most other rules, for the providers of AI systems, including those built on the pipeline |
| 2 December 2026 | New prohibitions on generating non-consensual intimate content and child sexual abuse material |
| 2 December 2027 | High-risk rules for Annex III systems (not uses we offer the pipeline for) |
Not done yet
- The documentation for builders on meeting Article 50 with the pipeline is planned, not written.
- We have not yet published the list of model providers the pipeline will route to.
- We hold no certification, and none exists yet for systems that are not high-risk.
Contact
Questions about this statement: contact@genesysbrain.com.