The PAN (Profiling Authorship Recognition) workshop is preparing for the 2026 edition with a focus on generative AI analysis and computational stylometry. The goal is to advance the field through objective and reproducible evaluations.

Planned Activities

PAN 2026 includes five main activities:

  1. Voight-Kampff Generative AI Detection: Detection of generative AI, especially in mixed and obfuscated authorship scenarios.
  2. Text Watermarking: A new task to identify and evaluate the robustness of existing text watermarking schemes.
  3. Multi-author Writing Style Analysis: Analysis of multi-author writing style, with the aim of identifying authorship change points.
  4. Generative Plagiarism Detection: Generative plagiarism detection, focused on source retrieval and alignment of generated text with original documents.
  5. Reasoning Trajectory Detection: Reasoning trajectory detection, with the goal of identifying the source and assessing the safety of reasoning trajectories generated by LLMs or written by humans.

Reproducibility and TIRA

As in previous editions, PAN invites software submissions in the form of easily reproducible Docker containers for most tasks. Since 2012, over 1,100 submissions have been made via the TIRA experimentation platform.