Project Conclusion December 2025
MISSION KI was completed in December 2025. This website documents the results of the project. The developed open-source tools, reports, and resources remain permanently available and can be used freely. Active maintenance or further development by the project team is no longer taking place. For any inquiries, acatech is happy to connect you with the right contacts: [kontakt@acatech.de]
MISSION KI was funded by the Federal Ministry for Digital Affairs and State Modernisation (BMDS).
Data Space Media
From Pilot Project to Sovereign European Media Infrastructure
The media sector is currently navigating a convergence of mutually reinforcing challenges. First, there is mounting pressure to decouple digital value creation from dominant platform ecosystems. Second, the imperative to use content and data responsibly in AI-driven processes—ranging from personalization to manipulation detection—is growing significantly. Third, the escalating threat of synthetic media and deepfakes underscores the urgent need for reliable verification mechanisms.
Consequently, the European Strategy for Data explicitly highlights the media sector as a target area for establishing data spaces to facilitate data access and reuse within a trusted European framework [1]. Concurrently, UNESCO warns of risks to information integrity that, absent robust technical and organizational countermeasures, threaten to undermine democratic processes [2]. On the economic front, UNESCO’s 2022/2025 World Trends Report, Journalism: Shaping a World at Peace, emphasizes that platform dependency and the consolidation of digital advertising and distribution power are eroding the bargaining power of traditional media players, thereby raising fundamental questions about digital sovereignty [3].
To counter these structural deficits in the media sector, MISSION KI launched a pilot project in 2025 in collaboration with leading public and private media organizations (ARD, ZDF, RTL Deutschland, ProSiebenSat.1, and the German Press Agency dpa). With acatech spearheading the effort, the initiative was backed by the Hamburg Ministry of Culture and Media, the co-initiators of the Beyond Platforms Initiative, and the Fraunhofer Institute for Software and Systems Engineering (ISST). The pilot aimed to deliver a proof of concept for a federated, decentralized data infrastructure. By October 2025, both the technical and organizational foundations had been laid, and technical viability was successfully demonstrated across five complex, prototypical use cases. The core finding: secure, privacy-compliant data exchange between competitors is not merely technically feasible, but economically advantageous.
The 4 Core Principles of the Media Data Space
For the Media Data Space to function as a trusted alternative to closed platforms, it is built on four foundational principles that were technically embedded during the pilot phase:
1. Data Sovereignty (Control via Policy)
In the platform economy, providers often lose control of their data the moment it is uploaded. The Media Data Space reverses this paradigm: the data provider retains absolute sovereignty. Every dataset is bound to a machine-readable contract (policy) that is technically enforced. The creator maintains granular control over who can access or even discover the data offerings, for what purpose (e.g., restricted to research, no commercial AI training), and for how long. Without an explicit opt-in, not a single byte is transferred.
2. Federated Decentralization (No Central Data Lake)
There is no monolithic central server storing Germany's media content. The architecture is strictly federated: data remains physically within the systems of the respective media organizations. Exchange occurs directly between partners via secure, encrypted peer-to-peer connections. This design eliminates single points of failure, bolsters resilience against cyberattacks, and mirrors the naturally decentralized structure of the media landscape.
3. Neutrality & Non-Discrimination
The operator of the data space infrastructure has no vested commercial interest in the content itself. As the Media Data Space scales in the future, the infrastructure could be operated by the acatech Foundation acting as a neutral trustee. This guarantees non-discriminatory access: a small startup or local publisher is granted the exact same technical access and conditions as a major broadcasting network. Algorithmic bias or preferential treatment by the operator is structurally precluded.
4. Interoperability & Open Standards
To actively dismantle vendor lock-in, the data space relies entirely on open-source software (Eclipse Dataspace Components) and open standards (IDSA, Dataspace Protocol, W3C ODRL). Consequently, participants have the freedom to choose—and switch—their technology providers at will. A participant using a connector from Vendor A today can migrate to Vendor B tomorrow without severing their connections within the ecosystem. This interoperability cultivates a vibrant, competitive market for technology service providers.
Deep Dive
The 5 Pilot Applications
During the pilot phase, the technical and organizational foundations were stress-tested across specific, real-world use cases. These five prototypical applications successfully delivered a proof of value, demonstrating the massive potential of the infrastructure:
1. Collaborative AI Training (GenAI)
The Challenge: Individual media organizations often lack the sheer volume and diversity of data required to train high-performing, German-language Large Language Models (LLMs) that meet rigorous journalistic standards.
The 2025 Solution: ARD, ZDF, and dpa virtually pooled their article archives via the data space to collaboratively develop an AI model for automated teaser generation. This leveraged the "Compute-to-Data" paradigm: the data never left the sovereign control of the respective publishers. Instead, the algorithm "traveled" to the data to train the model.
Key Insight: The AI model trained on this aggregated dataset delivered significantly better, lower-bias, and more precise teaser generation compared to isolated models, rapidly closing the gap with leading hyperscaler models.
Takeaway: Pre-competitive collaboration is the only viable path toward achieving competitive European AI sovereignty.
2. Federated News Verification
The Challenge: During major crises, hundreds of newsrooms often fact-check the exact same content (e.g., AI-generated deepfakes) in parallel. This results in a massive duplication of effort and wasted journalistic resources.
The 2025 Solution: The establishment of a shared metadata pool utilizing the C2PA (Content Credentials) standard. The moment a newsroom (e.g., dpa) verifies a video or debunks it as a fake, this status (the "verification stamp") is cryptographically secured and instantly shared across the data space.
Key Insight: During the testing phase, the redundant verification of identical disinformation was successfully eliminated. Response times to active disinformation campaigns improved measurably.
Takeaway: Massive efficiency gains through shared services directly bolster the resilience and defensibility of the democratic public sphere.
3. Privacy-Compliant Audience Segmentation
The Challenge: The deprecation of third-party cookies severely complicates targeted audience engagement—a void currently exploited by the "walled gardens" of US tech giants.
The 2025 Solution: Deploying a neutral technical trustee within the data space enabled the pooling of first-party data (e.g., interest profiles) in strict compliance with the GDPR. Data was evaluated in an encrypted, aggregated, and anonymized environment.
Key Insight: It is entirely possible to generate precise audience segmentation for advertising and content personalization without compromising user privacy or hemorrhaging data to US corporations.
Takeaway: A European counter-model to surveillance capitalism is technically viable.
4. Third-Party Service Integration (Breaking Vendor Lock-in)
The Challenge: Technically onboarding innovative third-party service providers (e.g., for personalization engines or paywall management) historically requires months-long integration projects.
The 2025 Solution: The publishing alliance DRIVE was integrated as a service via standardized data space connectors. Thanks to this standardization, onboarding functioned on a true "plug and play" basis.
Key Insight: The infrastructure drastically slashes transaction costs. Switching providers becomes seamless, which in turn invigorates market competition.
Takeaway: The architecture successfully mitigates dreaded "vendor lock-in" and fosters a dynamic, best-of-breed ecosystem.
5. Interoperability (The "Common Data Model")
The Challenge: Every media organization operates on disparate CMS platforms and proprietary data formats. This fragmentation has traditionally made technical collaboration prohibitively expensive and sluggish.
The 2025 Solution: The development of a "Common Data Model" for the media sector—essentially an Esperanto for media data. Connectors automatically translate between proprietary in-house formats and the universal data space standard.
Key Insight: Integration barriers for new partners plummeted. Complex new use cases could be deployed in a matter of weeks rather than months.
Takeaway: Standardization is the absolute, non-negotiable prerequisite for rapid scaling and broader market penetration.
ROADMAP 2026+
The Path to a Foundational European Infrastructure
With the successful conclusion of the pilot phase, the foundation has been set. The strategic roadmap for 2026 and 2027 outlines three critical phases to transition the initiative from a pilot project into full-scale production:
Phase 1: Institutionalization & Funding Acquisition (Q1/Q2 2026)
Securing public funding to subsidize the core infrastructure, directly addressing a recognized market failure. This phase also entails transitioning the project's operational structures into a neutral governance model under the umbrella of the acatech Foundation, strategically leveraging synergies with the Mobility Data Space.
Phase 2: Deployment & Onboarding (Starting Q3 2026)
The technical rollout of the foundational infrastructure—specifically Identity Management and the Clearing House—and the formal opening of the data space to broader market participation. A targeted focus will be placed on onboarding small and medium-sized publishers (SMEs) as well as tech startups to accurately reflect the full breadth and diversity of the market ecosystem.
Phase 3: Pan-European Integration (2027 and Beyond)
Federating with other European data spaces (such as the overarching EU Media Data Space and various Cultural Data Spaces). The ultimate objective is to establish a frictionless, cross-border single market for media data and AI models.
RESSOURCEN & PARTICIPATION
Ressourcen für Entwickler & Architekten
Full source code for the reference implementations:
Projektpartner
FREQUENTLY ASKED QUESTIONS (FAQ)
1. Has the project concluded?
The pilot phase (feasibility study and Proof of Concept) successfully wrapped up in October 2025. Currently (as of early 2026), the acatech Foundation is in the evaluation and strategic planning phase to establish the permanent infrastructure.
2. Why is public funding necessary?
While the participating media organizations initiated the development and contributed significant in-house resources, they represent only a fraction of the total market. To provide a truly neutral infrastructure accessible to everyone—including small local publishers and innovative tech startups—public subsidization of the foundational infrastructure is essential. This prevents the ecosystem from being monopolized by a few major players and dismantles prohibitively high barriers to entry.
3. What role does the acatech Foundation play?
It is positioned to act as the neutral operating entity and data trustee. This governance model would allow the Media Data Space to leverage the existing technological architecture of the Mobility Data Space. Doing so saves taxpayers millions in development costs for core components and unlocks powerful cross-sectoral synergies (e.g., in-car media consumption).
4. What happens next with the developed prototypes?
The five 2025 use cases serve as blueprints and reference implementations. In the upcoming phase, they will be hardened, transitioned into full-scale production, and opened up to additional participants, effectively evolving from prototypes into market-ready products.
5. How does this project differ from standard cloud providers?
The data space is not a cloud storage solution; it is a federated networking layer. It does not exclude cloud providers but rather integrates them—strictly on European terms. Users retain absolute data sovereignty and can seamlessly switch cloud providers without severing their established data connections within the ecosystem.