Thursday, November 19, 2026Thu, Nov 19 AboutRSSSubscribe

GDS

Design tools and creative technology

Design Tools

Google DeepMind Verifies Signal Encryption Using AI

Google's AI tool has proven Signal's encryption works, giving users confidence in the app's privacy guarantees. This development is significant as it shows AI can be used to improve cybersecurity.

Original reporting · zdnet.com

Original: “Google’s new AI tool just proved Signal encryption really works – here’s how”

Read “Google’s new AI tool just proved Signal encryption really works – here’s how” at ZDNET AI →
Google’s new AI tool just proved Signal encryption really works – here’s how
Image: ZDNET AI

How this was written. GDS wrote this article with a language model from ZDNET AI’s reporting (model: @cf/meta/llama-3.3-70b-instruct-fp8-fast); it describes that reporting rather than adding to it. The original, linked above, is the full story.

Key points from ZDNET AI’s report

  • Google DeepMind has used AI to verify parts of Signal's underlying code, giving users confidence in the app's privacy guarantees.
  • The use of AI in formal verification can accelerate the process and make it more efficient.
  • The researchers used a novel framework that combines human expertise with AI agents to achieve this milestone.
  • The development of AI-powered formal verification has the potential to improve the security infrastructure of widely deployed cryptographic protocols.
  • AI can be both a risk and a solution in cybersecurity, and companies are developing AI models specifically geared towards cybersecurity.

Google DeepMind researchers have successfully used artificial intelligence to verify parts of Signal's underlying code, according to ZDNET AI. This means they have mathematically proven that certain components of the encrypted messaging app work as intended without errors. The verification process is crucial in ensuring the security of communications protected by Signal's cryptography.

The use of AI in formal verification is a significant development, as it can accelerate the process and make it more efficient. Previously, manual verification was a time-consuming and labor-intensive task that required human expertise. The researchers used a novel framework that combines human expertise with AI agents to achieve this milestone.

The verification process involved dividing the task into two parts: humans identifying what needed to be proven, and AI agents executing the proof. The AI agents also audited the proof and highlighted areas that required further review by humans. This approach resulted in a significant time savings, with only 1% of the codebase requiring a second look.

The researchers tested their AI tool by using it to verify two previously fixed Signal bugs, and it produced correct results. This demonstrates the effectiveness of the tool in distinguishing between correct and flawed code. The successful verification of Signal's encryption is a major breakthrough, as it gives users greater confidence in the app's ability to deliver on its privacy guarantees.

The development of AI-powered formal verification has the potential to improve the security infrastructure of widely deployed cryptographic protocols. Google DeepMind's achievement marks the first time AI has been used to formally verify a widely deployed protocol at this depth. The researchers believe this is a step towards improved security infrastructure, especially at a time when AI is disrupting the cybersecurity space.

The breakthrough highlights the dual role of AI in cybersecurity: it can be both a risk and a solution. While AI has been involved in several security incidents, companies are also developing AI models specifically geared towards cybersecurity. The use of AI in formal verification is a significant step forward in improving the security of encrypted messaging apps like Signal.

The future of AI-powered formal verification looks promising, with companies like Anthropic, Google, and OpenAI developing models that can help improve cybersecurity. The year 2026 has seen several AI-related security incidents, but it has also seen significant developments in AI-powered cybersecurity solutions. As AI continues to evolve, it is likely to play an increasingly important role in improving the security of widely deployed cryptographic protocols.

Questions and answers

What is formal verification?

Formal verification is the highest standard of security assurance, which involves mathematically proving that software works as intended without errors.

How did Google DeepMind verify Signal's encryption?

Google DeepMind used a novel framework that combines human expertise with AI agents to verify parts of Signal's underlying code.

What are the implications of this development?

The development of AI-powered formal verification has the potential to improve the security infrastructure of widely deployed cryptographic protocols, giving users greater confidence in the privacy guarantees of encrypted messaging apps like Signal.

Read the full story at ZDNET AI More from Wednesday, November 11, 2026’s edition