Google Expands Gemini 4 Carbon Access After AI Agents Formally Verify Signal Protocol Code

Google has broadened access to its Gemini 4 Carbon model, allowing paying developers, businesses, and Google AI Ultra subscribers to use it through the Gemini API and Gemini for business. The expansion follows the successful use of Carbon-based AI agents to formally verify core security properties of the Signal protocol implementation in libsignal. Google DeepMind also introduced AlphaProof Code, which combines human experts with AI agents to generate millions of lines of Lean proofs that confirm guarantees such as forward secrecy and post-compromise security at the machine-code level. The verification effort, completed in four weeks by five researchers, also demonstrated the method's ability to detect previously patched flaws when they were reintroduced. In parallel, Carbon for cyber helped Wiz discover zero-days at two major SaaS providers and assisted Google's Cloud Vulnerability Research team in confirming the exploitability of two bugs in a popular open-source project. Google is releasing the specifications and proofs publicly while noting that the work applies to its own build of the code rather than Signal's production applications.

SecurityWeekTranslated from English

Read at source

Google on Tuesday expanded access to Gemini 4 Carbon and said agents built on the model helped formally verify core security guarantees in the code behind the Signal protocol. Paying developers and businesses can now use Carbon via the Gemini API and Gemini for business, and Google AI Ultra subscribers are getting access.

Introductory rates of $2/$10 per million input/output tokens apply until November 30, then double. Until now, only select cyber defenders in Google’s Fairwind Program could use Carbon. They still need Fairwind membership for Carbon for cyber, which Google says tops key benchmarks for discovering, exploiting and fixing vulnerabilities.

According to Google, Wiz used Carbon for cyber to find zero-days at two major SaaS providers within two hours. Google’s Cloud Vulnerability Research (CVR) team used the model to confirm that two bugs in a popular open source project were exploitable, with severity comparable to PHP2Shell, a pair of PHP flaws that attackers have already exploited.

Separately, Google DeepMind introduced AlphaProof Code, which pairs human experts with AI agents to formally verify software, mathematically demonstrating that it meets defined requirements. The team applied it to libsignal, the Signal protocol’s reference implementation. Versions of the protocol also run in WhatsApp and Google Messages.

Humans worked with agents running on Carbon and Claude Opus 5.5 to write the specifications defining what had to be proven. Carbon-based agents then generated all of the roughly 6.5 million lines of Lean proofs, with Lean rejecting any invalid ones. The proven guarantees, including forward secrecy, post-compromise security, and authentication, hold for the compiled machine code, while most protocol verification examines only the design.

Five researchers did the proof work in four weeks, something Google says has traditionally taken specialists years. To show the method catches real flaws, the team reinserted two old, already-patched libsignal bugs and proved that those builds break the security guarantees. The results cover Google’s own build of the code, run through a small command-line wrapper, and “not Signal’s own applications.” Side channels were also excluded from parts of the analysis.

Meridian digestThe week's most consequential security stories, distilled into one briefing.See the digest

More in AI & Model Risk

CISO Forum Virtual Summit 2026 Opens Today with Focus on AI Governance and Enterprise Defense
WhiteMatter Raises $20 Million in Series A Funding to Secure Enterprise AI Agents
Anthropic Releases Mythos 5.2 AI Model with Doubled Vulnerability Discovery in Firefox and Reduced Sandbox Escapes