alpha0 sp. z o. o. is implementing project FENG.02.12-IP.02-0458/26, co-financed by European Funds, entitled “Preparation of a Eurogrant application planned under the Horizon Europe programme”.

What is the purpose of the project?

The project aims to prepare and submit a Eurogrant application under Horizon Europe, for call HORIZON-CL3-2026-02-CS-ECCC-02 – “Enhancing the Security, Privacy and Robustness of AI Models and Systems (SecureAI)”. Preparatory activities include the application documentation, specialised analyses carried out by external contractors, expert consultations and necessary translations.

What will the planned Eurogrant cover?

The planned Eurogrant concerns the development of new methods, an architecture and a demonstrator for securing the entire lifecycle of artificial intelligence models and systems. The solution will support provenance tracking for data, models and components; automated integrity checks; detection of unauthorised changes, vulnerable dependencies and manipulated inputs; and secure transfer and deployment of models in offline and on-premises environments.

What is the purpose of the planned R&D work?

The R&D work will design and experimentally validate new mechanisms for improving the integrity, privacy and robustness of AI models. Research will examine how to link AI artefacts to their provenance, detect inconsistencies in data and dependencies, verify conditions for secure deployment, and produce auditable evidence of compliance without unnecessarily disclosing data or model parameters. A threat model, experimental scenarios and a research demonstrator will be developed.

Scope of the planned Eurogrant

The planned Eurogrant will in particular involve:

• developing a threat model and reference architecture for a security layer covering AI data, models and components;
• developing methods for provenance tracking and recording the change history of AI artefacts;
• designing automated validation of model, data and dependency integrity, including detection of unauthorised modifications;
• developing mechanisms to identify vulnerable or unverified components and manipulated input data;
• defining secure methods for transferring models between environments and deploying them offline and on-premises;
• producing auditable evidence of provenance and consistency while protecting privacy;
• building a research demonstrator and developing integrity-violation scenarios and security tests;
• preparing a validation plan for two or three pilots, with metrics for change detection, integrity, privacy and security overhead;
• assessing protection effectiveness, result reproducibility, deployment impact, and computational and memory overhead.

What will the planned Eurogrant deliver?

The planned Eurogrant is expected to deliver a reference architecture, threat model, specifications for AI artefact provenance tracking and integrity verification, a software demonstrator, validation scenarios and procedures, security metrics, and functional, security and performance test reports.

What will the project deliver?

The outcome of project FENG.02.12-IP.02-0458/26 will be the preparation of application documentation and a Eurogrant proposal for Horizon Europe, supported by specialised analyses used to define the planned R&D activities. Preparatory work also covers necessary translations, expert consultations and, where required, application supplements.

The documentation will enable the company to seek funding for R&D into methods of safeguarding AI model integrity and privacy throughout the AI lifecycle.

The direct beneficiary of the project results will be alpha0 sp. z o. o., which will use the documentation to seek funding for research and development on securing the lifecycle of AI models.

Project funding