All articles tagged with #confidential computing
Signal founder Moxie Marlinspike unveils Confer, an end-to-end encrypted, open-source AI chatbot that encrypts prompts on-device and runs inference in a confidential computing environment to keep conversations private and away from data brokers or training pipelines.
The adoption of generative AI poses significant security and privacy concerns for enterprises, as it can potentially expose proprietary data and lead to breaches of intellectual property. To address these challenges, confidential computing, a new approach to data security that protects data while in use and ensures code integrity, is emerging as a solution. Confidential computing isolates data and intellectual property from infrastructure owners and makes them accessible only to trusted applications running on trusted CPUs, ensuring data privacy through encryption. This technology can help enterprises embrace the power of generative AI without compromising on safety and compliance with evolving regulations.
Google Cloud and Intel collaborated on a nine-month audit of Intel's new hardware security product, Trust Domain Extensions (TDX), which revealed 10 confirmed vulnerabilities, including two significant ones. The review and fixes were completed before the production of Intel's fourth-generation Intel Xeon processors, which incorporate TDX. The project is part of Google Cloud's Confidential Computing initiative, and the goal is to help chipmakers find and fix vulnerabilities before they create potential exposure for Google Cloud customers or anyone else. Additionally, Google worked with Intel to open source the TDX firmware, low-level code that coordinates between hardware and software.