Goal will use it to summarize your conversations

Goal prepare new and controversial changes in WhatsApp. Recently announced that the messaging application would begin to show adsfor the first time in its history, in the sections of States and Channels. Now has advanced an update that introduces a little more AI in the user experience, raising new doubts about the privacy of WhatsApp -whose use, by the way, has just banned the US cybersecurity office Message Summaries either Message summaries.

If at any time you look over by the amount of unrewn messages, you can ask AI to generate a quick summary to understand the essentials of a conversation. To address the immediate concerns about the privacy that this generates, Meta affirms that it will use its technology of Private Processing either Private processing. Supposedly, it guarantees that neither goal neither WhatsApp can see your original messages nor the private summaries generated by AI.

Message summaries on WhatsApp.Goal.

To demonstrate that we can all be calm with that the AI ​​accessed your conversations, goal has published a technical document in which it explains how it is intended to maintain that privacy. In said Paper, Meta points out that the system is based on what it calls Trusted Execution Environment (TEE) either Reliable execution environments.

These TEE They are simplifiably, Safe and independent black boxes that work on hardware -driven servers such as NVIDIA H100 GPU. Your phone establishes a direct and encrypted connection with one of these Tee, sends the message data for processing and receives the summary. Supposedly, Neither Meta administrators nor other systems can access what happens within this box. In addition, once the process is finished, No message is preserved.

The goal system uses a process called Remote Attestation either Remote certification. Before sending any data, your phone requires a test to the server: a Cryptographic Report signed by the hardware itself that verifies that the TEE is executing exactly the software that target has publicly registered before a third as Cloudflare. Only if that verification is valid, the connection is established.

Users will have control over this function. It will be deactivated by defaultand from the section Advanced Chat Privacy It can be specified what conversations can be shared with the functions of AI. Message Summaries It is already beginning to be implemented in English for users in the US global expansion Throughout next year.