Paper Accepted Then Rejected: Research Use of Sky Sports Commentary Videos and Consent Issues
Analysis
This situation highlights a significant challenge in AI research involving publicly available video data. The core issue revolves around the balance between academic freedom, the use of public data for non-training purposes, and individual privacy rights. The journal's late request for consent, after acceptance, is unusual and raises questions about their initial review process. While the researchers didn't redistribute the original videos or train models on them, the extraction of gaze information could be interpreted as processing personal data, triggering consent requirements. The open-sourcing of extracted frames, even without full videos, further complicates the matter. This case underscores the need for clearer guidelines regarding the use of publicly available video data in AI research, especially when dealing with identifiable individuals.
Key Takeaways
- •Consent requirements for using public broadcast footage in research are not always standardized and can vary by journal.
- •Extracting and processing personal data (e.g., gaze information) from videos, even without redistribution, can trigger consent requirements.
- •Researchers should clarify data usage and consent requirements with journals *before* submitting papers to avoid unexpected rejections.
“After 8–9 months of rigorous review, the paper was accepted. However, after acceptance, we received an email from the editor stating that we now need written consent from every individual appearing in the commentary videos, explicitly addressed to Springer Nature.”