AI-CODE - AI services for COntinuous trust in emerging Digital Environments

Projektinformationen

AI-CODE

ID Finanzhilfevereinbarung: 101135437

DOI

10.3030/101135437

EK-Unterschriftsdatum 20 Oktober 2023

Startdatum 1 Dezember 2023

Enddatum 30 November 2026

Finanziert unter

• Digital, Industry and Space

Gesamtkosten

Keine Daten

EU-Beitrag

€ 4 969 471,25

Koordiniert durch

DS TECH SRL
Italy

Periodic Reporting for period 1 - AI-CODE (AI-CODE - AI services for COntinuous trust in emerging Digital Environments)

Berichtszeitraum: 2023-12-01 bis 2024-11-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

The media sector is undergoing rapid transformation, driven by unprecedented technological advancements that continuously reshape the way information is created, distributed, and consumed. Among these innovations, artificial intelligence (AI) has emerged as a pivotal force, influencing the media landscape both positively and negatively. This challenge is further compounded by the rise of next-generation social media platforms, including decentralized and immersive environments. These emerging platforms, while fostering novel opportunities for interaction and content sharing, also introduce new vulnerabilities, such as complex content manipulation techniques. These challenges necessitate urgent action and innovative solutions to uphold media freedom and pluralism, ensuring the delivery of trustworthy information. In this context, initiatives such as the European Democracy Action Plan and the strengthened Code of Practice on Disinformation, part of the Digital Services Act, emphasize the importance of fostering technological advancements that protect and empower the media sector.
AI-CODE aims to evolve state-of-the-art research to achieve a novel set of six services that are both content and users-based aligning well with the vision above by addressing the dual challenges of combating disinformation through the detection of manipulated information and the creation of transparent, credible and trusted content and by coaching media professionals in understanding evolving digital environments and leveraging AI technologies.
Its mission is to enable a sustainable transformation of the media ecosystem, ensuring that AI serves as a tool for positive societal impact in creating reliable content and more trustful digital environments.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

AI-CODE project achieved significant advancements during its Study phase. Key outcomes include an in-depth analysis of emerging trends in social media platforms, such as federated platforms like Mastodon and Bluesky and immersive virtual environments. These analyses highlighted both promising opportunities and challenges for AI-CODE services. While immersive environments remain predominantly gaming-focused with younger audiences, federated platforms demonstrated potential for combating disinformation, albeit with unique obstacles to address. A comprehensive benchmarking of state-of-the-art technologies was conducted to position AI-CODE’s offerings effectively and address unmet needs in the media sector. Methodologies for data collection from decentralized social media were developed, enabling the reconstruction of social network dynamics and the analysis of disinformation and hate speech propagation on federated platforms.
Generative AI research explored both its risks and benefits. This included assessing Gen-AI’s ability to generate disinformation, such as synthetic visual media and textual narratives in low-resource languages like Greek. Cutting-edge detection techniques for synthetic content were benchmarked across 22 languages, leading to the development of innovative datasets and methods for tasks such as fallacy detection, check-worthy claims identification, and low-resource text classification using large language models (LLMs). Generative AI was also evaluated for its role in enhancing content production workflows, including paraphrasing and summarization, while addressing authorship concerns to ensure trusted information production.
A user-centered approach defined target user groups for each service, developed use scenarios, and established early functional and non-functional requirements alongside key ethical considerations. Additionally, early mock-ups and prototypes were designed based on feedback from co-creation workshops, ensuring alignment with user needs and fostering collaborative development.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

The knowledge collected and developed on next-generation social media and the dual role of AI in generating both manipulated and trusted content will be leveraged to advance state-of-the-art research outcomes. These outcomes will provide tools designed to guide and coach media professionals in producing more trusted information, incorporating the effective use of generative AI and the new online environments introduced by next-generation social media. Media professionals stand to benefit significantly from AI-CODE’s advanced AI-based services, which will address three key modalities—text, image, and video—while supporting text processing in both high-resource and low-resource European languages and providing a set of advanced features.
These services are expected to have a transformative impact on their daily workflows by automating time-intensive tasks, such as detecting disinformation or identifying credible content, and by utilizing generative AI to enhance content production. AI-CODE aims to facilitate the adoption of AI, particularly generative AI, across the full spectrum of information production processes. Its universal applicability ensures relevance during major societal events, such as elections, as well as during crises, including pandemics, conflicts, and wars. All communicable results from the AI-CODE project will be disseminated to the European R&D community using standardized, recognized, and widely adopted channels, ensuring maximum accessibility and impact.