At Amazon’s recent Fall hardware event, executive David Limp announced that the company would be enhancing the capabilities of its voice assistant, Alexa, using generative AI technology. This move follows the introduction of AlexaLLM, a large language model designed to improve the assistant’s responses. However, it has now been revealed that Amazon is also using user conversations to train Alexa’s AI capabilities. While this may not come as a surprise given the data practices of other tech companies, it serves as a reminder of the importance of user data to corporations and the limited privacy protections in place.
According to NBC News, Amazon has confirmed that it is using user conversations with Alexa for product improvement and development. An Amazon spokesperson defended this practice, stating that training Alexa with real-world requests is essential for delivering an improved experience. They also claimed that only a small fraction of user voice recordings are used in this process, although exact numbers were not provided. The company assured users that they still have access to privacy controls and tools to manage their Alexa experience.
This recent news is not the first privacy concern surrounding Amazon Alexa. In the past, it was discovered that Amazon employees were able to listen to and transcribe conversations had with the assistant. This led to the development of a feature that allows users to stop human screeners from accessing their voice commands. Users who are concerned about their privacy can adjust Alexa’s privacy settings or choose to remove their Alexa devices altogether.