Signal’s generator helps with Meta AI coding

moxy marlinspike, Privacy advocate who created the secure calling app signal A widely used open source encryption protocol, He said This week, a privacy-focused AI platform, Grantedwill begin integrating its technology into Meta’s AI systems.

Every day, billions of chat messages sent via Signal, WhatsApp’s Meta, and Apple Messages are protected End-to-end encryption. This feature, which makes it impossible for tech companies and anyone other than the sender and recipient to snoop on your messages, has become prevalent over the past decade. With the growing popularity of generative AI platforms, people are now also exchanging billions of messages daily using AI-powered chatbots that don’t offer end-to-end encryption protection — making it easier for AI companies to access what you’re talking about.

This is by design, as platforms often want to train their AI models on as much user data as possible and they have succeeded in doing so. It’s hard to pull out of Use your information as training data. But like chatbots and Artificial intelligence agents Some have become more capable Technicians Companies are pushing to create more restrictive and privacy-focused systems.

“As LLMs continue to be able to do more, we should expect more data to flow to them,” Marlinspike wrote in an article. Short blog post About his cooperation with Meta published on Tuesday. “Right now, none of this data is private. It is shared with AI companies and their employees, hackers, subpoenas, and governments. As is always the case with unencrypted data, it will inevitably end up in the wrong hands.”

Marlinspike wrote that it “will integrate Confer’s privacy technology to support Meta AI.” He also emphasized that grants, which For the first time At the beginning of this year, it will continue to operate independently from Meta. Marlinspike added that the project’s goal is to provide technology that “allows everyone to have the full power of artificial intelligence along with the complete privacy of an encrypted conversation.”

In 2016, Marlinspike worked with WhatsApp, which is owned by Meta, to roll out end-to-end encryption to more than one user. Billion accounts simultaneously. Over the past year, WhatsApp has introduced a Meta AI chatbot into its app, which it is not Protected by the company In the same way as individual conversations.

“People use AI in ways that are highly personal and require access to confidential information,” said WhatsApp chief Will Cathcart books On Wednesday on the social media platform X about the cooperation with Confer. “It’s important that we build this technology in a way that gives people the ability to do this privately.”

The adoption of crypto AI is still emerging. The cryptographic systems used in end-to-end encryption of traditional digital communications cannot be easily or directly translated into data protection for generative AI. For its part, Confer is still a new project, and Marlinspike’s blog post did not provide specific details about how the collaboration with Meta will work or the specific goals of the integration.

Neither Marlinspike nor Meta provided WIRED with any additional comment prior to publication.

“It would be great for people who use chatbots that use Meta AI to have anonymity and privacy within that exchange,” says Mallory Knodel, a crypto researcher at New York University. More importantly, this means that Meta will not be able to access AI-powered chat data for training purposes, says Knodel, who and colleagues recently published research Study on end-to-end cryptography and artificial intelligence. “I really hope more AI chatbots adopt this approach.”

Knodel’s initial and initial reviews of Confer Pointing The platform isn’t perfect, but it’s an important example of how to create your own AI-powered chatbot.