Seminal
The magazine and atlas of EU-funded researchA new publication from the web and AI team at Scott Charnley

Atlas / Germany

European Research Council grant · Running

Privacy4FMs Privacy Protection and Auditing for Foundation Models

Coordinator
Cispa - Helmholtz-Zentrum Fur Informationssicherheit Ggmbh, Germany
Runs
1 January 2026 to 31 December 2030
Status
Running
EU funding
€1.5m
Total cost
€1.5m
Partners
1 organisations in 1 countries
Type
European Research Council grant
Grant agreement
101220235

What the project is doing

In the project's own words, from its summary on CORDIS.

Novel foundation models (FMs) like GPT, LLaMA, and Stable Diffusion are achieving exceptional performance across diverse tasks, generating high-quality text, images, and audio, and driving industry innovations. This progress stems from a shift in machine learning paradigm: instead of training task-specific models on curated datasets, FMs are first pretrained on vast, uncurated data to become strong general-purpose models, then adapted on smaller, domain-specific datasets for specific tasks.

However, FMs leak information from their training data. For example, recent studies reveal that they can re-create individual data points from their pretraining and adaptation datasets. This poses serious privacy risks when private data is involved. Preventing exposure requires developing methods to ensure privacy-preservation throughout FMs' lifecycle, from pretraining to deployment. To achieve this, our project will identify sources of privacy leakage, provide privacy guarantees over both pretraining and adaptation, and audit FMs to detect privacy violations.

Therefore, we must overcome three major challenges: the limited understanding of privacy risks in FM pretraining, the lack of formal joint privacy guarantees for pretraining and adaptation, and the ineffectiveness of current privacy auditing methods. The solution that we propose will establish a novel theoretical framework for privacy guarantees in FMs under the pretrain-adapt paradigm. Our fundamental innovations rely on the insight that, due to complex interdependencies between pretraining and adaptation data, different data points require individual levels of protection to prevent leakage.

Advancing methods for identifying, achieving, and accounting for such individual guarantees will enable us to formally bound privacy leakage over both training stages and to detect violations. These innovations will allow society to benefit from technological advancements through FMs without compromising individuals' privacy.

Topics

Who is involved

OrganisationCityCountryRole
Cispa - Helmholtz-Zentrum Fur Informationssicherheit GgmbhSaarbruckenGermanyCoordinator

Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI