DeepMind researchers said they formally verified parts of the Signal protocol
Google DeepMind researchers described a four-week research project in which, they said, an AI framework called AlphaProof Code was used to formally verify a number of the Signal protocol's key cryptographic guarantees. The account comes from WIRED, in a report dated November 10, 2026, at 12:01 AM, and from Axios, in a report published the same day. The claims below are attributed to what those outlets reported from DeepMind researchers, not to a DeepMind or Google company blog post.
Real, reported: based on the public sources cited below. Nothing added or dramatized. Sources checked November 10, 2026, at 7:14 AM ET (EST). This is a company research claim reported by journalists. Neither WIRED nor Axios described the work as peer-reviewed or named a scientific journal, and we do not do so here.
What the researchers said they did
According to WIRED, the DeepMind analysis formally verified a number of the Signal protocol's key cryptographic guarantees, using agentic AI, as a proof-of-concept completed in four weeks. Axios, citing DeepMind, reported that the agents were powered by the company's Gemini 4 Carbon model and generated more than 3 million lines of mathematical proofs. That model name and the 3-million-line figure appear in Axios, not in WIRED, and should be read as Axios's report of what DeepMind said.
WIRED reports that the analysis did not find any new vulnerabilities in the portions of the Signal protocol it looked at. When the researchers used AlphaProof Code on older versions of Signal that contain known and long-fixed vulnerabilities, WIRED says, it correctly identified them and showed that the protocol could not be formally verified. Axios similarly reported that the researchers reintroduced two previously fixed bugs into Signal's code and that, in both cases, the framework generated mathematical proofs showing that the altered software was insecure.
WIRED also reports that the researchers are releasing the specification and proof they developed for the Signal protocol analysis, to aid other formal verification work. We have not reviewed those materials.
Limits the sources themselves emphasize
The same WIRED article is careful about scope. It says the analysis "was limited in scope—it did not, for example, formally verify the entire Signal mobile app." DeepMind researchers, WIRED writes, emphasize that their findings are just one milestone in a larger process. Swarat Chaudhuri, a DeepMind research scientist, told WIRED: "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."
WIRED reports that the DeepMind researchers point out that they were able to get a head start to a degree because of some past formal verification work that the Signal Foundation has conducted. Formal verification, as WIRED explains it, uses math to check that software is functioning as intended. WIRED notes that, before the AI era, it was incredibly time-consuming for people to describe the characteristics of software they wanted to prove, turn those summaries into formulas, and then develop the proof, and that agentic AI can significantly cut down the time of the latter project. That is a method for checking stated properties of specified code. It is not, by itself, a demonstration that every related application, deployment, or future code change is free of flaws. WIRED also writes that AI-aided formal verification is having a moment, but that it is unclear how practical it will be for widespread use and how effective it can be at securing large amounts of code.
Axios reported Chaudhuri saying, "You can finally get to this world of unhackable software." That is a researcher's quoted aspiration, not a finding that Signal, or software in general, is unhackable. Axios also reported that the latest feat remains a research project, not a commercially available cybersecurity product. WIRED says Google declined to share the cost of the analysis and declined to comment on whether it has plans to release a commercial formal verification product or service.
How the outlets describe the significance
WIRED frames the work in two ways: as further validation of cryptography used in communications on platforms such as the Signal app, Google Messages and WhatsApp, and as a demonstration of DeepMind's formal verification capabilities with AlphaProof Code. Ehren Kret, Signal's chief technology officer, said in a statement quoted by WIRED: "While prior formal verification focused on smaller portions of the code, Google DeepMind's approach enables significant expansion. As AI-driven security threats grow, rigorously verifying the Signal protocol implementation gives us even greater confidence in our code's privacy guarantees." That is a statement from Signal about work on Signal, reported by WIRED. It is not a peer-reviewed evaluation, and it is not an independent laboratory audit we have seen.
Pushmeet Kohli, described by WIRED as chief scientist at Google Cloud and DeepMind's vice president of research, spoke about a longer-term hope of verifying implementations so that organizations that rely on assurances could say, in his words as quoted by WIRED, "I can prove these properties." WIRED itself calls it an open question whether AI-automated formal verification can truly scale, and at a workable cost. Chaudhuri, as quoted by WIRED, also warned against a "vibe world where AI is generating the goals and AI is generating the proofs," and described a human-AI collaboration process that still seeks human input and review at key junctures.
Our advice (general guidance)
Readers should not treat this coverage as proof that the Signal app as a whole, or any messaging product that uses related cryptography, has been mathematically shown to be free of every possible defect. WIRED's report limited the claim to portions of the protocol examined in a research project. Axios framed the result more broadly, reporting that researchers used AI to prove that the encryption code protecting billions of private messages works as intended and to verify that the code underpinning Signal's encryption is secure. A formal proof of specified properties is only as meaningful as the specification, the portion of the code in scope, and the human review around it.
If you use Signal or another messaging app, keep using official app stores and official update channels. This research report is not a reason to ignore ordinary account-security habits, and it is not a reason to trust unsolicited messages that claim "Signal has been proven unhackable" or that ask you to move a conversation onto another app or website. DeepMind's project, as reported, is not a consumer security product you can turn on, and Axios reported that it is not commercially available.
These are editorial precautions, not individualized legal, cryptographic or security advice.
Sources
- WIRED: A New AI Tool From Google DeepMind Proved the Signal Protocol Is Secure (Nov 10, 2026)
- Axios: Google touts AI breakthrough, path to \"unhackable\" software (Nov 10, 2026)
Published November 10, 2026 by FraudMeNot Team. Label: Real, reported.