Google DeepMind uses agentic AI to formally verify the security of the Signal protocol
Google DeepMind has completed a formal verification of the Signal protocol, mathematically proving that the protocol satisfies its security claims under all conditions. The protocol is also used by WhatsApp and Google Messages; the verification targets the protocol itself rather than the Signal app as a whole. When testing older versions, the method identified known vulnerabilities and flagged them as failing verification. DeepMind research scientist Swarat Chaudhuri said there is still a long way to go before this method can be extended to all software worldwide, and that the process is extremely compute-intensive, with it remaining unclear whether it can be used more broadly.
Why read this
AI score 70
DeepMind has completed a formal verification of the Signal protocol using agentic AI; readers can learn about this method's progress at the intersection of cryptography and AI, as well as its compute limitations.
Topics
Body
The Signal protocol has long been considered the gold standard for end-to-end encrypted messaging, with Edward Snowden a notable advocate. Now a Google DeepMind analysis has “formally verified” the security of the protocol.
In addition to its use in the Signal app, the same protocol is also used by WhatsApp and Google Messages, and the work has been described as “a landmark development” …
“Formal verification” is the term given to mathematically proving or disproving that a cryptographic algorithm or protocol satisfies its security claims under all conditions. It guarantees that there is no situation in which the protocol will fail. This is a far higher standard than traditional testing methods which seek to find vulnerabilities but cannot promise that others do not exist.
Using agentic AI to carry out formal verification is a huge breakthrough for cryptography and AI alike. It is, however, extremely computationally demanding, so Google told Wired it’s unclear as yet how practical it will be to use it more broadly.
DeepMind researchers emphasize that their findings are just one milestone in a larger process. “We are very excited by this result, but there is a long distance to go before we are able to scale this to all the software of the world, and we want to do that,” Swarat Chaudhuri, a DeepMind research scientist, tells WIRED.
It’s important to note that the validation is for the protocol itself rather than the Signal app as a whole. However, encouragement can be taken from the fact that when tested against older versions of Signal that contained known vulnerabilities, it identified those and flagged them as failing formal verification.
Also worth stating is that Google was able to build on some past formal verification work already completed without the use of AI. This analysis took four weeks, and the scale of resources required has not yet been disclosed, so we shouldn’t expect a whole array of other protocols to be validated overnight. All the same, this is a very big deal.
meross Smart Garage Door Opener
Wireless CarPlay adapter
Logitech MX Master 4 for Mac