The ALLIES project aims at supporting micro (and small) HSPs in achieving compliance with the new requirements and obligations under the TCO Regulation. To achieve this, ALLIES adopts four main approaches, each targeting skill development and innovation. These four pillars will be respectively focused on (1) learning and awareness raising, (2) technical development and adaptation, (3) training and education, and (4) experience sharing and reporting mechanisms. Both the learning and awareness raising initiatives will further expand the capacity of the TCO Regulation by designing a taxonomy of online terrorist related behaviours. The second pillar will be focused on the development of a specific set of AI-boosted user-friendly tools that will enhance small HSPs capabilities on more accurately recognising and removing suspicious content. The training and education aspects focus will not only rest on the TCO Regulation content, but also on familiarisation activities with the developed tools. The fourth approach aims at creating a safe online environment where end users anonymously share their experiences, as well as useful data in addition to a risk assessment module and unified reporting form for terrorist-related content and behaviour. Complementing and cooperating with each other, these four pillars will not only support HSPs, but will also contribute to the fulfilment of the TCO Regulation objectives.

MKLab is responsible for scientific and technical management of the ALLIES project and has a critical role in the project by leading a WP on Terrorist-related data acquisition and processing, aiming to develop, test, and deploy a specific set of AI boosted modular tools as well as to create a safe online environment for experience and data sharing. Furthermore, MKLab also leads the development of appropriate algorithms for providing multimodal hash representations of the content hosted by the ALLIES HSPs aiming to facilitate the inexpensive detection of identical or highly similar material posted on the same or different providers to enable the quick correlation among hashes to find similar content, whilst ensuring that reverse engineering cannot be applied to recreate the original material.





  • Spathi Theoni
  • Kalpakis George
  • Tsikrika Theodora
  • Vrochidis Stefanos