11-07-2025
The European Research Council (ERC) has awarded a prestigious Proof of Concept (PoC) grant to Joana Gonçalves-Sá, researcher at NOVA LINCS, the research centre of NOVA School of Science and Technology | NOVA FCT, and at LIP (Laboratory of Instrumentation and Experimental Particle Physics), to lead the project FARE_DSA – A Framework for Auditing Recommendation Engines under the DSA.
FARE_DSA will develop and test an innovative methodology to audit algorithmic platforms that are increasingly crucial in shaping how citizens access information online — particularly search engines and large language models (LLMs). It will be the first project to provide a concrete, policy-aligned tool for implementing Article 40 of the European Union’s Digital Services Act (DSA), which grants qualified researchers access to non-public platform data to assess systemic risks to the EU, such as electoral interference or political manipulation.
“Platforms like Google or ChatGPT have become true gateways to information,” says Joana Gonçalves-Sá. “But we are still far from understanding how their algorithms influence public opinion. During the 2024 European Parliament elections, we observed that both search engines and LLMs systematically displayed content biased toward one side of the political spectrum. With this project, we aim to test whether these biases are widespread and, if so, understand why. This can only be achieved through collaboration with both regulators and the platforms themselves, so we can identify causes and study their implications.”
The project builds on previous ERC-funded research, including the ERC Starting Grant FARE – Fake News and Real People (launched in 2020 with €1.5 million), and the PoC grant FARE_Audit, which enabled the development of customizable online “robots” (web crawlers) capable of mimicking human behaviour and simulating different characteristics (e.g. country, age, gender). These bots interact with search engines or LLMs and collect data on how these platforms respond to their queries. Analyses using this approach have shown that small differences in browsing history or user profile can lead to significantly different — and sometimes biased — results.
FARE_DSA will expand this methodology by accessing real platform data under the DSA, offering new insights into how personalisation and monetisation may distort information access, especially during critical periods such as elections.
“Our team has spent years developing ethical, privacy-preserving tools to audit algorithmic systems without collecting personal data,” adds Joana Gonçalves-Sá. “But to truly understand what happens behind the scenes — for example, how search results are ranked or why certain links appear first — we need access to non-public data. Article 40 enables exactly that. This is a unique moment for research in algorithmic transparency. This funding will allow us to open that door — and, we hope, help others follow through.”
The project will also help clarify which data is available, in what formats, and under what conditions it can be accessed under Article 40, helping to resolve a long-standing impasse between researchers and online platforms. FARE_DSA will also generate public policy recommendations, organise training sessions for academics and NGOs, and contribute to other audits of very large online platforms and search engines (VLOPs and VLOSEs).
Although ERC PoC grants are typically associated with commercial innovation, FARE_DSA demonstrates its potential as a high-impact social innovation tool, bridging scientific excellence, digital transparency, and public policy implementation.
The project was awarded €150,000 and will run for one year.