AI Providers Address Metadata Leak Threatening Encrypted Chats

Microsoft disclosed a new side-channel attack named 'Whisper Leak' that can allow eavesdroppers to infer chat topics despite end-to-end encryption. Major AI chatbot providers, including OpenAI, Microsoft, Mistral, and xAI, have implemented defenses to safeguard user privacy. The attack exploits patterns in encrypted packet sizes and timing during streaming responses.