- Juli 29, 2026
- --
Magnolia DXP and the EU AI Act
-
Laura DelnevoProduct Owner, Magnolia
Key insights
Magnolia DXP does not build, train, or operate its own AI models. It connects to third-party providers — such as OpenAI, Anthropic, and AWS Bedrock — through its own AI Connector framework.
Magnolia DXP does not rebrand these models as its own. The provider behind any AI capability in Magnolia DXP is always a named third party, operating under its own regulatory obligations.
Every AI-assisted content action is tracked in a full audit log: what changed, when, which model was invoked, and by which user.
Magnolia's AI capabilities are not developed, marketed, or intended for use as high-risk AI systems under Article 6 of the EU AI Act — including Annex III use cases like employment decisions, creditworthiness assessment, or access to essential services.
Article 50 of the EU AI Act introduces transparency obligations for AI-generated content, applicable from August 2, 2026. Magnolia DXP is building toward the specific measures the Act requires.
Magnolia DXP and the EU AI Act
What this means for you
If you use Magnolia's agentic AI, the AI capabilities you rely on run on models from established third-party providers, connected through Magnolia DXP's own governed integration layer — not a Magnolia-built or Magnolia-trained model. That distinction matters for how you think about your own compliance obligations, though it doesn't remove them: as the organization deploying AI-assisted content to your own users, you remain responsible for how you communicate that to your teams and your audience, in line with your obligations under Article 50.
Magnolia DXP's role is to give you the infrastructure to meet that responsibility: an audit trail you can export for regulatory reporting, and transparency controls you can configure to meet your own disclosure obligations.
Article 50: What we’re building
Article 50 introduces transparency requirements for providers and deployers of certain AI systems, applicable from August 2, 2026. Magnolia DXP is building toward:
Machine-readable provenance metadata for AI-generated images and content, consistent with C2PA/IPTC standards (Article 50.2).
Customer-configurable disclosure labels for AI-generated content on published pages (Article 50.4).
Accessibility-compliant implementation of all disclosure mechanisms (Article 50.5).
Not built or intended for high-risk use
Magnolia's AI capabilities are not developed, marketed, or intended for use as high-risk AI systems within the meaning of Article 6 of the EU AI Act — including the use cases listed in Annex III, such as employment and workers management, creditworthiness assessment, or access to essential services, or as a safety component of products covered by Annex I. Whether a specific deployment falls within a regulated use case is each customer's own assessment to make.
Due diligence stays with you
Magnolia DXP's agentic AI is an extensible, composable platform. We aim to provide built-in components — field-level labels and icons marking AI-assisted or AI-powered content — that surface AI use in the authoring interface. Implementing these correctly in your live environment is your responsibility, or your implementation partner's. If your teams deploy AI-assisted content to your own users, you're accountable for communicating that clearly, in line with your own obligations under the Act.
Read the full statement
This post summarizes the key points. For Magnolia's complete, legally-reviewed position, read the EU AI Act Statement.
Nothing in this post should be interpreted as legal advice. Compliance with the EU AI Act and other applicable regulations depends on your specific implementation, use case, and governance measures.
