This intensive program at PJAIT (Warsaw) combines data science and psychology to train future researchers in combating disinformation. Participants will take part in lectures and team projects using real data, guided by experts from Poland and abroad. The program culminates in a supervised data analysis challenge and a publication workshop. The school is held in conjunction with the AI Summit PJAIT 2026 (Sept. 16), which participants will attend to connect theory with current AI practice. NAWA co-funds participation, and partial scholarships are available.

Key Dates:

• Applications Open: May 20, 2026
• Deadline (online): June 20, 2026, 11:59 p.m. CEST
• Notification of Acceptance: June 30, 2026
• School Dates: September 12–18, 2026 (AI Summit on September 16)

Organizers:

A team at the Polish-Japanese Academy of Information Technology (PJAIT), in partnership with the University of Koblenz, the Open University of Cyprus, NAWA, and the EUonAIR University Alliance.

Contacts

For inquiries:

Kinga Skorupska (kinga.skorupska@pjwstk.edu.pl).
Adam Wierzbicki (adam.wierzbicki@pjwstk.edu.pl)
Arkadiusz Modzelewski (arkadiusz.modzelewski@pja.edu.pl)

Topics: ICT & Psychology Concepts Related to Disinformation Detection and Debunking

The International Summer School on Disinformation Detection and Debunking is a doctoral-level, research-oriented training program that combines:

1. computational methods for detecting, explaining, and mitigating disinformation,
2. cognitive-psychological mechanisms that drive the formation and persistence of beliefs, and 
3. responsible, evidence-based debunking practices. 

The curriculum is based on the “information disorder” framework (misinformation, disinformation, and malinformation) and on empirical findings that misinformation is often resistant to correction without well-designed interventions.

The course is taught in English and organized into two closely related modules. 

The ICT block covers: data science workflows; machine learning and deep learning for credibility assessment; NLP for stance, evidence, and persuasion signals; multimodal analysis (image/video + text); and human-centered interface design for analyst decision support. 

The Psychology/Ethics block covers: the foundations of cognitive psychology, perception and attention, and FATE (fairness, accountability, transparency, and ethics) as they apply to disinformation detection systems.

The central learning component is a supervised dataset challenge. Participants work in small interdisciplinary teams on curated, license-compliant datasets such as MIPD (Modzelewski et al., 2024),  MALINT (Modzelewski et al., 2026), or ISOT Fake News (Ahmed et al., 2018), with at least one multilingual and one multigenre track, and with explicit attention to manipulation techniques and malicious intent in disinformation. Teams will submit reproducible pipelines and short research reports. The top projects (based on leaderboard rankings and scientific review) will enter a post-school mentorship track with lecturers to develop publishable outputs

Educational rationale and focus on best practices

Research on disinformation and misinformation shows that “corrections” can fail unless they are designed with careful consideration of cognition, narrative coherence, and trust; this is why the school explicitly combines detection with debunking and human factors. The program also treats adversarial behavior as a primary concern (e.g., manipulation techniques, robustness), reflecting current evaluation practices in shared tasks that emphasize robustness and adversarial examples in credibility-related domains.

Goals

1. enable participants to design and evaluate disinformation-detection systems under realistic constraints (domain shift, persuasion techniques, and synthetic content);
2. connecting algorithmic signals to psychological mechanisms and debunking strategies; 
3. build lasting international collaborations among PhD researchers and practitioners.

The supervised dataset challenge will begin with an introduction to key datasets that will serve as the foundation for participants’ preliminary research projects. These will include datasets developed by the program lecturers, such as MIPD (Modzelewski et al., 2024) and MALINT (Modzelewski et al., 2026), as well as widely used and highly cited public resources in natural language processing, such as the ISOT Fake News dataset (Ahmed et al., 2018). The MIPD and MALINT datasets have been manually annotated by experienced disinformation researchers and fact-checking experts with at least 3–5 years of professional experience, ensuring high-quality and reliable labels.

Following the introductory session, participants will be divided into teams to identify and refine research questions to be explored during the summer school. This phase will include guided brainstorming sessions with lecturers, as well as a review of relevant academic literature.

Each team will then conduct a preliminary research project based on the selected datasets. The outcomes of this work will include a structured presentation of the research problem, methodology, and initial findings, as well as an outline of a potential research paper.

At the end of the summer school, teams will present their work, and both the presentations and research proposals will be evaluated by the instructors.

• Arkadiusz Modzelewski, Giovanni Da San Martino, Pavel Savov, Magdalena Anna Wilczyńska, and Adam Wierzbicki. 2024. MIPD: Exploring Manipulation and Intention in a Novel Corpus of Polish Disinformation. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pp. 19769–19785, Miami, Florida, USA. Association for Computational Linguistics.
• Arkadiusz Modzelewski, Witold Sosnowski, Eleni Papadopulos, Elisa Sartori, Tiziano Labruna, Giovanni Da San Martino, and Adam Wierzbicki. 2026. MALicious INTent Dataset and Inoculating LLMs for Enhanced Disinformation Detection. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3125–3148, Rabat, Morocco. Association for Computational Linguistics.
• Ahmed, H., Traore, I., & Saad, S. (2018). Detecting opinion spam and fake news using text classification. Security and Privacy, 1(1), e9.