We would like to inform all participants of an issue that occurred during the preparation of the MRIxFields2026 dataset. Due to an unintended operational error during the release of the Google Drive dataset, the paired ground-truth data for 20 validation subjects (contained in Training_prospective.zip) were temporarily accessible for a period of time before being identified and immediately removed. We sincerely apologize for this mistake and any inconvenience it may have caused. To ensure the fairness and integrity of the challenge, all teams are strictly prohibited from using the withdrawn validation data in any stage of model development. If your team has downloaded these data, please permanently delete them immediately. The withdrawn paired validation data must not be used for any of the following purposes, including but not limited to: ● Pre-training ● Model training ● Validation ● Early stopping ● Hyperparameter tuning ● Model selection ● Pseudo-label generation ● Data augmentation ● Any direct or indirect optimization using the paired ground-truth information To restore fairness for all participants, we have established a Data Integrity Policy, which can be found here: https://www.synapse.org/Synapse:syn72060672/wiki/642237 The key requirements are summarized below: ● Complete training logs are mandatory. Teams must submit complete training logs together with their final Docker submission. The logs should demonstrate that the withdrawn validation data (and any data derived from them) were never accessed during any training epoch. ● Full transparency of evaluation results. During the final Docker evaluation stage, the results on the training, validation, and hidden test datasets will all be released to participants to facilitate transparency and community verification. ● Mandatory reproducibility package. Together with the final Docker submission, teams must also submit the complete reproducible training code. The Organizing Committee will independently reproduce the submitted methods for integrity verification. These measures are introduced solely to ensure a fair competition for every participant. We understand that they impose additional work on participating teams, and we sincerely appreciate your understanding and cooperation. We also welcome community oversight. If any participant identifies credible evidence of potential violations, please contact the Organizing Committee. All reports will be handled confidentially and investigated carefully. Finally, we would like to emphasize that the withdrawn dataset refers only to the paired validation ground-truth data contained in Training_prospective.zip that were temporarily available on Google Drive. The previously released unpaired validation data remain part of the official challenge dataset and may continue to be used according to the challenge rules. We sincerely apologize again for this incident and thank you for your understanding, professionalism, and continued support of MRIxFields2026. We remain committed to maintaining a fair, transparent, and reproducible challenge for the entire community.