Call for Interest: ChildSafeAds @ NLLP (EMNLP 2026)
Children are exposed to commercial content on YouTube every day, yet detecting, categorising, and assessing compliance with advertising regulations remains an open challenge.
We are excited to announce ChildSafeAds, a new shared task co-located with the NLLP Workshop at EMNLP 2026.
Interested in participating? Register your interest: https://forms.office.com/e/q0JSkSuGzD
ChildSafeAds is the first NLP benchmark to jointly address:
Detection and categorisation of commercial content
Compliance assessment with EU advertising law
Retrieval-augmented reasoning over legal sources
The task uses a large-scale dataset of community-verified sponsored segments from SponsorBlock. Evaluation will take place on CodaBench, and selected teams will be invited to submit a system description paper for publication in the ACL Anthology (subject to peer review).
Shared tasks:
ST1 — Commercial Type: Classify the legal nature of promotions (e.g., physical goods, digital content, services, digital services, or no commercial content).
ST2 — Product Category: Multi-label classification across categories including toys, food, apps, fashion, health, education, financial products, and more.
ST3 — Compliance Risk Flags: Retrieval-augmented classification over legal sources (EU directives, CJEU case law, and platform guidelines) to identify risks such as undisclosed advertising, direct exhortation, prohibited products, or misleading claims.
Timeline:
20 July 2026 — Training and development data release
3 August 2026 — Evaluation phase opens
11 August 2026 — System submission deadline
1 September 2026 — System description paper deadline (invited teams)
15 September 2026 — Notification
22 September 2026 — Camera-ready papers
28 October 2026 — NLLP @ EMNLP 2026
The full call for participation, including dataset details, evaluation protocol, and submission instructions, will follow soon.
If you are interested in NLP, computational social science, AI for public interest, or legal NLP, we would love to have you participate.
Questions: Thales Bertaglia (t.f.costabertaglia [at] uu.nl)
Organising team: Thales Bertaglia (Utrecht University), Catalina Goanta (Utrecht University), Güneş Acar (Radboud University), Gerasimos (Jerry) Spanakis (Maastricht University)