Periodic Reporting for period 1 - ATHENA (An exposition on THe forEign informatioN mAnipulation and interference)
Período documentado: 2023-11-01 hasta 2025-01-31
Russia, China and other countries have exploited the Internet to wage campaigns of disinformation and interference in Europe to disrupt democratic processes for their perceived political and economic benefit. Since its invasion of Ukraine, Russia has greatly increased its cyber-attacks and disinformation campaigns against Ukraine and other countries in Europe. ATHENA is contributing to Europe’s defence against foreign information manipulation and interference (FIMI). It has examined 32 manifestations of FIMI in case studies, including the tactics, techniques and procedures (TTPs) used by attackers. The consortium has analysed the behavioural and societal effects of FIMI. It is assessing and developing countermeasures against TTPs. It is collecting and processing FIMI data, developing a knowledge graph, a disinformation analysis toolbox and a dashboard showing the prevalence of FIMI and the countermeasures in action. ATHENA is conducting a comparative, eight-country analysis of legal and regulatory frameworks against FIMI and their shortcomings and where Member States and the EU should strengthen their policy framework. ATHENA will produce an easy-to-use questionnaire that anyone can use to decide whether they are confronted by FIMI. We are going beyond state of the art by scanning the horizon for new instances of FIMI, for signs of the evolution in disinformation (as with deepfakes and chatbots). From the start of the project, ATHENA has been engaging stakeholders through interviews, workshops, a Stakeholder Board and a cluster of relevant projects. We are sharing knowledge with other stakeholders and will provide training in countermeasures to universities and schools.
Work package 1: Analysis of FIMI campaigns and techniques
The consortium has already met Objective 1 of analysing manifestations of FIMI, the TTPs used by attackers and their impacts on democracy. We completed our analysis of 32 case studies of FIMI (two more than promised in the Grant Agreement) and collected primary data via 20 interviews to gather views on FIMI, the role of technology in shaping FIMI and its overall impact on democratic processes.The partners have examined the psychological mechanisms, persuasive tactics and societal consequences of FIMI. The partners analysed the psychological mechanisms, persuasive tactics, and societal consequences employed by malign actors to manipulate perceptions, undermine trust, and destabilise democratic institutions.
Work package 2: Countermeasure techniques – state of the art and beyond
The partners conducted a literature review of the existing empirical research of countermeasures. We developed a taxonomy based on existing classifications of countermeasures. We are planning a methodology to measure their effectiveness.
Work package 3: Sourcing, storing, engineering and exploring data
We developed our initial scraping framework for harvesting FIMI data found online. We are defining a harmonized data model for capturing harvested data in our knowledge graph. The partners have developed a knowledge graph for capturing FIMI data. We have also coordinated with the DISARM Foundation to investigate uplifting the current DISARM framework to a semantic representation. A prototype ontology was submitted to the DISARM Foundation in June 2024 and is still under active development. We have identified tools for deployment including Deepfake Detection, Entity Recognition, News Reliability Classifier, Fake News Detection, Fake News Categorization, Hate Speech Detection, and an LLM-based assistant for querying the knowledge graph. The partners have a dashboard for using the tools. We have identified functional and non-functional requirements for end users. We organised an initial requirements gathering workshop in October 2024 with expert analysts investigating FIMI to assess their major pain points with existing tools.
Work package 4: Policymaking and foresight
The partners have analysed the current legal landscape at an EU level, with attention given to the AI Act and the DSA. In the coming months, this analysis will expand by looking into the Media Freedom Act and several others, as well as gaining insight into unique differences between member states. The consortium completed a detailed draft evaluating emerging technologies most likely to impact disinformation in the next five years. We submitted an interim version of our report, which is not due until Oct 026. Work on developing a questionnaire for evaluating potential FIMI will start in M18. Preliminary planning has been already started. The consortium will make recommendations within EU policy priorities.
The consortium has already met Objective 1 of analysing manifestations of FIMI, the TTPs used by attackers and their impacts on democracy. We completed our analysis of 32 case studies of FIMI (two more than promised in the Grant Agreement) and collected primary data via 20 interviews to gather views on FIMI, the role of technology in shaping FIMI and its overall impact on democratic processes.The partners have examined the psychological mechanisms, persuasive tactics and societal consequences of FIMI. The partners analysed the psychological mechanisms, persuasive tactics, and societal consequences employed by malign actors to manipulate perceptions, undermine trust, and destabilise democratic institutions.
Work package 2: Countermeasure techniques – state of the art and beyond
The partners conducted a literature review of the existing empirical research of countermeasures. We developed a taxonomy based on existing classifications of countermeasures. We are planning a methodology to measure their effectiveness.
Work package 3: Sourcing, storing, engineering and exploring data
We developed our initial scraping framework for harvesting FIMI data found online. We are defining a harmonized data model for capturing harvested data in our knowledge graph. The partners have developed a knowledge graph for capturing FIMI data. We have also coordinated with the DISARM Foundation to investigate uplifting the current DISARM framework to a semantic representation. A prototype ontology was submitted to the DISARM Foundation in June 2024 and is still under active development. We have identified tools for deployment including Deepfake Detection, Entity Recognition, News Reliability Classifier, Fake News Detection, Fake News Categorization, Hate Speech Detection, and an LLM-based assistant for querying the knowledge graph. The partners have a dashboard for using the tools. We have identified functional and non-functional requirements for end users. We organised an initial requirements gathering workshop in October 2024 with expert analysts investigating FIMI to assess their major pain points with existing tools.
Work package 4: Policymaking and foresight
The partners have analysed the current legal landscape at an EU level, with attention given to the AI Act and the DSA. In the coming months, this analysis will expand by looking into the Media Freedom Act and several others, as well as gaining insight into unique differences between member states. The consortium completed a detailed draft evaluating emerging technologies most likely to impact disinformation in the next five years. We submitted an interim version of our report, which is not due until Oct 026. Work on developing a questionnaire for evaluating potential FIMI will start in M18. Preliminary planning has been already started. The consortium will make recommendations within EU policy priorities.
· ATHENA is publishing the first book on FIMI. It is the first detailed analysis of so many case studies. It provides a more detailed analysis of the TTPs used by attackers than exists elsewhere. It provides a comprehensive and up-to-date analysis of deepfake technologies.
· It has developed a new taxonomy of countermeasures. Our taxonomy groups all these together under 10 new headings that gives a wider overview than any current taxonomy.
· ATHENA's knowledge graph has begun capturing vast quantities of news articles from websites linked to FIMI attacks. The scope and scale of this graph will provide an ever-growing source of insights into how FIMI attacks are executed.
· ATHENA submitted an interim report on horizon scanning in March. It is the first detailed horizon scan of technologies likely to be used for disinformation in the next three to five years. It is the first time there has been a match-up between these technologies and the TTPs.
The partners have all the tools we need to fulfil the project’s objectives.
· It has developed a new taxonomy of countermeasures. Our taxonomy groups all these together under 10 new headings that gives a wider overview than any current taxonomy.
· ATHENA's knowledge graph has begun capturing vast quantities of news articles from websites linked to FIMI attacks. The scope and scale of this graph will provide an ever-growing source of insights into how FIMI attacks are executed.
· ATHENA submitted an interim report on horizon scanning in March. It is the first detailed horizon scan of technologies likely to be used for disinformation in the next three to five years. It is the first time there has been a match-up between these technologies and the TTPs.
The partners have all the tools we need to fulfil the project’s objectives.