Challenge

The workshop will host a community challenge on the rapid evolving task of Synthetic Image Source Attribution. In particular, it will be based on a newly ad hoc released attribution dataset, comprising state-of-the-art text-to-image generators. The competition will invite participants to carry out closed and open set attribution experiments, as well as data processing tasks.

Key Guidelines:

📣 The Challenge will start on March 9th, 2026.

🗓️ Participants can work on their solutions until May 29th 2026. 

⏰ Paper Submission deadline: June 6th 2026 

📝 Participants may submit a conference paper through the regular submission procedure (CMT) to be considered for publication in conference proceedings. Alternatively, proposals can be submitted as a technical report* (min 2-4 pages, not included in the conference proceedings) to the workshop email address: deepfake.icann26work@gmail.com. In both cases, the code must be publicly available to ensure reproducibility and allow verifications by the organizers.

🏅Winners will be presented at the workshop session. 

*Please, use the following LaTeX template for the technical report. [download]

Quick Rules:

• the dataset is released on the Kaggle platform, which will host and handle the challenge [link];

• participants may compete in one or more of the following tasks: closed-set attribution, open-set attribution, dataset processing;

• participants may compete alone or in teams;

• eligibility for the official workshop ranking is restricted to submission of technical report/conference paper + code, and registration to the conference (of at least one participant).

For the complete regulation, please refer to the following [document].