Privacy-preserving human-AI dialogue systems – Organisation of a technological challenge
EDF-2025-LS-RA-CHALLENGE-DIGIT-HAIDO
Funding Amount
Amount not specified
Deadline
16/10/2025
Eligible Organization Types
Non-profit, Private Business, Research Institution, Public Institution
Trusted by
What is the Privacy-preserving human-AI dialogue systems – Organisation of a technological challenge Call?
Grant Description
This call aims to establish a common testing environment and organize a technological challenge to evaluate human–artificial intelligence dialogue systems for defence use cases, addressing the lack of standard metrics and trustworthiness. It supports data collection, annotation, distribution, the definition of clear evaluation protocols and metrics, performance measurement, and debriefing workshops with defence users to produce shared databases and foster a European defence dialogue community.
Required Results for Successful Funding
Expected Outcomes
Funded projects are expected to deliver:
• A standardised, reproducible testing environment for human-AI dialogue systems in defence contexts.
• Enhanced clarity and transparency of system performance for developers, funders and end users.
• A vibrant European defence community around dialogue system evaluation, with sustained stakeholder engagement.
• Trustworthy dialogue systems that can handle classified information and justify decisions, thereby improving operational decision-making.
• Curated, annotated datasets and databases representative of defence use cases, made available for future research and development.
• Detailed, stakeholder-agreed evaluation plans, metrics and protocols, with objective performance measurements across multiple systems.
• Structured debriefing workshops and events to disseminate findings, gather feedback and guide next-generation challenge iterations.
• A governance framework and user-board model for ongoing steering and coordination of the challenge and subsequent activities.
• A standardised, reproducible testing environment for human-AI dialogue systems in defence contexts.
• Enhanced clarity and transparency of system performance for developers, funders and end users.
• A vibrant European defence community around dialogue system evaluation, with sustained stakeholder engagement.
• Trustworthy dialogue systems that can handle classified information and justify decisions, thereby improving operational decision-making.
• Curated, annotated datasets and databases representative of defence use cases, made available for future research and development.
• Detailed, stakeholder-agreed evaluation plans, metrics and protocols, with objective performance measurements across multiple systems.
• Structured debriefing workshops and events to disseminate findings, gather feedback and guide next-generation challenge iterations.
• A governance framework and user-board model for ongoing steering and coordination of the challenge and subsequent activities.
Funded areas
Scope Requirements
Proposals must organise a technological challenge on human-AI dialogue in accordance with the preliminary evaluation plan (Annex 4) and shall include:
• Data lifecycle management: collection, annotation, quality assessment, curation and distribution of datasets reflective of defence scenarios.
• Evaluation framework: development of detailed plans, metrics, rules, tools and significance tests for measuring system performance, iterated through draft reviews with participating teams.
• Test campaign planning: specification of use-case scenarios, languages covered, data annotation volume and representativeness, plus a detailed schedule (timeline/Gantt) of all challenge phases.
• Performance measurement: objective execution of test campaigns under agreed protocols and systematic recording of results.
• User engagement: establishment of a defence user board, involvement in live system tests, and structured feedback mechanisms.
• Iterative refinement: submission of draft evaluation plans early to incorporate feedback; any evolution of plans must consider technical feasibility, cost, scientific relevance and military representativeness.
• Governance model: clear description of steering bodies, roles, responsibilities and decision-making processes at challenge level.
• Openness: challenge must be open to teams funded under EDF-2025-LS-RA-CHALLENGE-DIGIT-HAIDP and other sources.
• Data lifecycle management: collection, annotation, quality assessment, curation and distribution of datasets reflective of defence scenarios.
• Evaluation framework: development of detailed plans, metrics, rules, tools and significance tests for measuring system performance, iterated through draft reviews with participating teams.
• Test campaign planning: specification of use-case scenarios, languages covered, data annotation volume and representativeness, plus a detailed schedule (timeline/Gantt) of all challenge phases.
• Performance measurement: objective execution of test campaigns under agreed protocols and systematic recording of results.
• User engagement: establishment of a defence user board, involvement in live system tests, and structured feedback mechanisms.
• Iterative refinement: submission of draft evaluation plans early to incorporate feedback; any evolution of plans must consider technical feasibility, cost, scientific relevance and military representativeness.
• Governance model: clear description of steering bodies, roles, responsibilities and decision-making processes at challenge level.
• Openness: challenge must be open to teams funded under EDF-2025-LS-RA-CHALLENGE-DIGIT-HAIDP and other sources.
Additional Conditions for Applicants
Special Conditions
- Single-stage lump-sum grant (EDF-LS): opening 18 February 2025; deadline 16 October 2025, 17:00 Brussels time.
• Admissibility: proposals must respect page limits and layout (sec.5 of call document; Part B template) and submit Part B and annexes as a password-protected zip (AES-256).
• Eligibility: participants and affiliated entities from eligible countries (sec.6); financial and operational capacity checks (sec.7); exclusion criteria apply.
• Evaluation & award: processes per sec.8; award criteria, scoring and thresholds per sec.9; indicative evaluation and grant-agreement timeline per sec.4.
• Legal/financial set-up: follow sec.10, EDF Regulation 2021/697, PSI, EU Financial Regulation; use model grant agreement EDF-AG-LS.
• Mandatory coverage: proposals must at minimum deliver all "integrating knowledge" tasks; activities outside art.10(3)(a–d) (e.g. prototyping, testing, certification) are not eligible.
• Budget constraints: funding requests must stay within the topic’s indicative lump-sum budgets (€7 M for DIGIT-HAIDO; €20 M for HAIDP-STEP).
• Documentation: include Application Form (Part A), Part B, Detailed Budget Table, Participant Info, declarations (ownership/control, PRS, indirect costs), and Annex 4 preliminary evaluation plan.
• Admissibility: proposals must respect page limits and layout (sec.5 of call document; Part B template) and submit Part B and annexes as a password-protected zip (AES-256).
• Eligibility: participants and affiliated entities from eligible countries (sec.6); financial and operational capacity checks (sec.7); exclusion criteria apply.
• Evaluation & award: processes per sec.8; award criteria, scoring and thresholds per sec.9; indicative evaluation and grant-agreement timeline per sec.4.
• Legal/financial set-up: follow sec.10, EDF Regulation 2021/697, PSI, EU Financial Regulation; use model grant agreement EDF-AG-LS.
• Mandatory coverage: proposals must at minimum deliver all "integrating knowledge" tasks; activities outside art.10(3)(a–d) (e.g. prototyping, testing, certification) are not eligible.
• Budget constraints: funding requests must stay within the topic’s indicative lump-sum budgets (€7 M for DIGIT-HAIDO; €20 M for HAIDP-STEP).
• Documentation: include Application Form (Part A), Part B, Detailed Budget Table, Participant Info, declarations (ownership/control, PRS, indirect costs), and Annex 4 preliminary evaluation plan.
Important dates
Open from: 18/02/2025
Deadline:16/10/2025
Win your grant with Cogrant
How cogrant works
Smart Funding Fit Check
Instantly know if your project fits call criteria. Our system highlights strengths, gaps, and risks, so you only apply when your chances are real.
Intelligent Research Engine
Our research engine digs deeper than any founder could alone. It pulls data about your company, your market, and even your competitors - giving your proposal an edge backed by facts, not fluff.
Full Proposal Development
Our system builds your entire proposal from start to finish, ensuring every section is complete, coherent, and crafted to the highest quality. You focus on your vision — we deliver the winning document.
Expert Review
Our experts run a complete review before delivery, ensuring your draft proposal is accurate, well-structured, and polished. You’ll receive a high-quality draft that’s ready for your final touches before submission.
Our Numbers
Results that speak for themselves
70+
proposals prepared
€36.7M
total funding secured for clients
65%
win rate in highly competitive calls