SafetyPublished All news from that day
Google DeepMind reports AlphaProof Code agents proved Signal protocol security properties in four weeks.
DeepMind says AlphaProof Code, run with Gemini Carbon, produced machine-checked proofs of key security properties.
Check our sources · 6 facts from 1 sourceGoogle DeepMind says AI agents proved key security properties of the code behind the Signal protocol in four weeks, a task it says has historically taken expert teams years. The claim is in a DeepMind blog post of 10 November 2026 describing a framework it calls AlphaProof Code.
DeepMind says agents powered by Gemini Carbon generated more than six million lines of Lean, a language for machine-checked proofs, and that human experts reviewed every formal specification. The agents then flagged one percent of the codebase for a second human look.
To test the method, DeepMind says it reintroduced two real, previously fixed bugs into an isolated copy of Signal's code, and in both cases the system produced a machine-checked proof that the buggy version was insecure.
Signal's chief technology officer, Ehren Kret, is quoted in the post calling the use of AI agents to translate live code into valid Lean theorems a major breakthrough. DeepMind says it is open-sourcing its formal verification artifacts.
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against this source.
1 Introducing AlphaProof Code (10 Nov 2026)
Open the source archived copy-
Google DeepMind's post of 10 November 2026 says its researchers "guided AI agents to mathematically prove key security properties of the actual code underlying the Signal protocol" using "AlphaProof Code, a novel agentic framework for formal verification".
citeguided AI agents to mathematically prove key security properties of the actual code underlying the Signal protocol ... AlphaProof Code, a novel agentic framework for formal verification
-
DeepMind says: "Even at this complexity, our verification process took just four weeks – a task that historically takes expert teams years".
citeour verification process took just four weeks - a task that historically takes expert teams years
-
DeepMind says its researchers translated Signal's security goals into formal mathematical specifications "with human experts reviewing every one", and that "agents powered by Gemini Carbon did the proofs, generating over six million lines of Lean".
citewith human experts reviewing every one ... agents powered by Gemini Carbon did the proofs, generating over six million lines of Lean
-
DeepMind says the agents audited the proof and highlighted areas for a second human look, "just one percent of the codebase".
citejust one percent of the codebase
-
DeepMind says that to confirm AlphaProof Code can tell correct code from buggy code, "we reintroduced two real, previously fixed bugs into an isolated version of Signal's code. In both cases, the system produced a machine-checked proof that the buggy version was insecure."
citewe reintroduced two real, previously fixed bugs into an isolated version of Signal's code. In both cases, the system produced a machine-checked proof that the buggy version was insecure.
-
The post quotes Ehren Kret, Chief Technology Officer of Signal: "Using AI agents to translate live code into valid Lean theorems is a major breakthrough." and says "We are open-sourcing our formal verification artifacts for community use."
citeUsing AI agents to translate live code into valid Lean theorems is a major breakthrough. ... We are open-sourcing our formal verification artifacts for community use.
Every sentence above is checked against this source. Each fact has its own address you can cite.
Source 1 · Google DeepMind · 10 Nov 2026
Introducing AlphaProof Code (10 Nov 2026)
Google DeepMind's post of 10 November 2026 says its researchers "guided AI agents to mathematically prove key security properties of the actual code underlying the Signal protocol" using "AlphaProof Code, a novel agentic framework for formal verification".
guided AI agents to mathematically prove key security properties of the actual code underlying the Signal protocol ... AlphaProof Code, a novel agentic framework for formal verification
DeepMind says: "Even at this complexity, our verification process took just four weeks – a task that historically takes expert teams years".
our verification process took just four weeks - a task that historically takes expert teams years
DeepMind says its researchers translated Signal's security goals into formal mathematical specifications "with human experts reviewing every one", and that "agents powered by Gemini Carbon did the proofs, generating over six million lines of Lean".
with human experts reviewing every one ... agents powered by Gemini Carbon did the proofs, generating over six million lines of Lean
Show all 6 factsShow fewer facts
DeepMind says the agents audited the proof and highlighted areas for a second human look, "just one percent of the codebase".
just one percent of the codebase
DeepMind says that to confirm AlphaProof Code can tell correct code from buggy code, "we reintroduced two real, previously fixed bugs into an isolated version of Signal's code. In both cases, the system produced a machine-checked proof that the buggy version was insecure."
we reintroduced two real, previously fixed bugs into an isolated version of Signal's code. In both cases, the system produced a machine-checked proof that the buggy version was insecure.
The post quotes Ehren Kret, Chief Technology Officer of Signal: "Using AI agents to translate live code into valid Lean theorems is a major breakthrough." and says "We are open-sourcing our formal verification artifacts for community use."
Using AI agents to translate live code into valid Lean theorems is a major breakthrough. ... We are open-sourcing our formal verification artifacts for community use.
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep