AI regulation
Luxembourg opens TAID.LU call for a deepfake detector before EU AI Act rules bite
Companies have until 22 June 2026 to apply for a GovTech Lab innovation partnership that would give ALIA a secure tool for checking whether manipulated video, audio or images were properly disclosed.

Luxembourg is looking for a technical tool to help its media regulator decide when deepfake content has been objectively identified. The GovTech Lab opened the TAID.LU innovation partnership on 20 May 2026, with applications due by 22 June 2026 through the public procurement portal.
The project is aimed at the Autorité luxembourgeoise indépendante de l'audiovisuel, or ALIA. The government notice says the tool should support ALIA in its new supervision work linked to deepfake obligations under the EU Artificial Intelligence Act. The GovTech Lab call says ALIA may have to take administrative decisions on whether a deployer complied with a labelling obligation, and that such decisions cannot rest only on a state agent's visual inspection.
TAID.LU is therefore not framed as a public app for citizens. It is planned as a secure platform for ALIA agents, combining an interface, a processing engine and an AI layer that can detect suspected deepfakes and explain the result. The procurement notice identifies the procedure as an innovation partnership, a format used when a public buyer is seeking a solution that still needs to be developed.
The EU context matters because deepfake transparency is moving from policy debate into compliance work. The European Commission says the AI Act entered into force on 1 August 2024 and will be fully applicable two years later, with transparency rules coming into effect in August 2026. Article 50 of the Act requires deployers of AI systems that generate or manipulate image, audio or video content constituting a deep fake to disclose that the content was artificially generated or manipulated, subject to exceptions.
For Luxembourg media, political communication and online platforms, the important point is evidence. If ALIA is asked to assess whether disputed content should have been labelled, it needs a defensible method rather than a subjective screen-by-screen judgement. The call explicitly links that need to objectively verified facts and to Article 50(4) compliance or non-compliance.
Companies interested in TAID.LU should treat 22 June 2026 as the immediate deadline. The available public documents do not yet define a demo day, but they do make clear that the chosen solution must work in an administrative environment where explainability, traceability and secure handling of potentially sensitive audiovisual content will matter as much as detection performance.
Frequently asked
- What is TAID.LU?
- TAID.LU is a GovTech Lab innovation partnership seeking a secure tool to detect deepfake content for Luxembourg's audiovisual regulator, ALIA.
- What is the application deadline?
- The GovTech Lab page lists the public procurement period from 20 May 2026 to 22 June 2026.
- Why does ALIA need a deepfake detector?
- The call says ALIA may need to make administrative decisions about whether deepfake labelling obligations were respected, and those decisions need objectively verified facts rather than visual inspection alone.
Sources
Around Politics
A look at recent reporting on politics from the Étude newsroom.
Related by topic
Other Étude stories tagged with the same topics as this article.
More in Politics
Trending at Étude
Security policy Luxembourg maps a EUR1.67bn defence effort by 2029, while reserving judgment on NATO's 2035 target
Cost of living Luxembourg's 5.2% inflation reading: the fuel shock behind the number, and what it means for indexation
EU trade policy Europe tightens its steel defences — and the move lands closest to Luxembourg
Eurogroup Luxembourg at Eurogroup: digital euro, stablecoins and energy prices move onto the same agenda



