{"id":147923,"date":"2025-11-10T23:11:54","date_gmt":"2025-11-10T21:11:56","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/ataka-whisper-leak-po-opredeleniyu-tematiki-obshheniya-s-ai-chatbotami-cherez-analiz-tls-trafika"},"modified":"2025-11-10T23:11:54","modified_gmt":"2025-11-10T21:11:56","slug":"ataka-whisper-leak-po-opredeleniyu-tematiki-obshheniya-s-ai-chatbotami-cherez-analiz-tls-trafika","status":"publish","type":"post","link":"https:\/\/prohoster.info\/en\/blog\/news\/ataka-whisper-leak-po-opredeleniyu-tematiki-obshheniya-s-ai-chatbotami-cherez-analiz-tls-trafika","title":{"rendered":"The Whisper Leak attack involves determining the topic of conversation with AI chatbots through TLS traffic analysis.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
Microsoft security researchers developed a side-channel attack technique called Whisper Leak, which allows the classification of query topics to services based on large language models through passive analysis of encrypted traffic transmitted over TLS connections. Information about the size of network packets and the delays between their transmission proved sufficient to determine the topics of requests to AI chatbots with over 98% accuracy. In practice, the proposed method can be used to identify specific query topics in user transit traffic, such as attempts to obtain information about illegal activities, without decrypting the content. <\/p>\n
The toolkit for extracting data from traffic dumps, training models, and testing the method's functionality has been published on GitHub. The possibility of launching an attack has been demonstrated for 28 popular large language models from major providers. For instance, the accuracy of detecting queries on the topic of \"money laundering\" for many AI services reached 100%, with only 1 relevant query present in the analyzed traffic among 10,000 unrelated queries.
<\/center> <\/p>\n
The reason for the information leak is that models generate responses to requests step by step, one token at a time, using the previous token as context to determine the next most likely word or phrase. Consequently, each token is sent in a separate network packet, and the delay between the packets corresponds to the delay in the model's determination of the next token. <\/p>\n
In TLS, if data compression is not used, the ciphertext size equals the size of the plaintext plus a constant. By creating a model that correlates the desired token sets with packet sizes and delays between their transmissions, one can quite accurately identify the presence of the desired topics in the traffic. In the course of the study, three versions of such machine learning models were prepared, based on the architectures of neural networks LightGBM, Bi-LSTM, and BERT. Experiments were conducted for each model to determine the desired topic by analyzing either packet size, only delays between packets, or both criteria.
![\"The](\"\/wp-content\/uploads\/2025\/11\/9e2db8a564ae07b2ae7ced2d2f744e46.png\")
<\/center> <\/p>\n![\"The](\"\/wp-content\/uploads\/2025\/11\/d6be536d5b8814af45a42317c9454a44.png\")
<\/center> <\/p>\n