Research in conversation
Talks & media
Sharing research on machine unlearning, responsible AI, and information integrity.
Speaking
Talks & events
WIPE-OUT 2 / ECML PKDD 2026
Between Suppression and Collapse: Evaluating Narrative Unlearning with LENS
I presented our work with George Fletcher at the 2nd Workshop on Machine Unlearning and Privacy Preservation. LENS evaluates narrative unlearning by examining targeted suppression alongside output degradation.
AI HOUSE / Align, Forget, Repeat
Can we teach LLMs to forget, and why do we need to?
Чи можна навчити LLM забувати і навіщо це потрібно?
I spoke about machine unlearning: how it works, whether knowledge can be removed without losing a model’s capabilities, and why this matters for AI safety.
Memory Undone at Undone Computer Science
Conference participation and presentation of our work on knowing, not knowing, and responsible forgetting in data systems.
In the news
Media & institutional news
FairForget selected for Women Led Science support
The programme features our team’s research on fairness after machine unlearning.
Programme announcementPhD student Viktoriia Makovska took part in Undone Computer Science 2026 in Luxembourg
UCU Research, 2026.
Institutional news item about my participation in the Undone Computer Science 2026 conference in Luxembourg.
Ukrainian student created an AI model that detects Russian propaganda in Wikipedia
Texty.org.ua, July 31, 2025.
Coverage of my master's research prototype for detecting manipulative narratives in Ukrainian and Russian Wikipedia edits. The project is a research prototype, not a deployed Wikimedia moderation system.