[DeepSeek: Inputting certain characters triggering abnormal responses is due to model hallucinations caused by special characters, not related to security issues or privacy breaches] According to a report by Jinse Finance, on May 19, DeepSeek released a statement regarding abnormal responses triggered by special characters in the model. Recently, it was observed that some users reported receiving unexpected content when inputting such special characters during conversations with the DeepSeek model, leading some users to suspect 'conversation leaks.' After a comprehensive investigation, the technical team found that the abnormal responses triggered by inputting certain characters are due to model hallucinations caused by special characters and are not related to security issues or privacy breaches. DeepSeek stated that it will enhance the model's ability to recognize and handle special characters through targeted training, address related known issues, and optimize the model's performance in such scenarios.