Skip to content
NewA calmer, faster ZeroHour

Google widens Gemini 4 Carbon access after Signal verification

Original title: Google Widens Gemini 4 Carbon Access After AI Agents Verify Signal Code

SecurityWeekby Eduard Kovacs ()2 min read
Part of a story covered by 6 sourcesGoogle opens Gemini 4 Carbon to paying users at launch ratesMerged summary & timeline
AI summary · x-ai/grok-4.7Summaries by Vani AI

Paying users get API access at $2 and $10 per million tokens until November 30; Carbon for cyber stays in Fairwind.

Google expanded Gemini 4 Carbon to the Gemini API, Gemini for business, and Google AI Ultra subscribers, with introductory rates that double after November 30. DeepMind’s AlphaProof Code used human experts and agents on Carbon and Claude Opus 5.5 to formally verify libsignal. Carbon-based agents generated about 6.5 million lines of Lean proofs for forward secrecy, post-compromise security, and authentication on compiled machine code. Google is releasing the specifications and proofs; the result covers its own build through a command-line wrapper, not Signal’s applications.

Unconfirmed: Claims that Carbon tops cyber benchmarks and that Wiz found zero-days in two hours come from Google.

Key facts
  • Until now only selected Fairwind Program defenders could use Carbon.
  • Google said five researchers completed the proof work in four weeks.
  • Reinserted, already-patched libsignal bugs were shown to break the guarantees.
  • Some side-channel analysis was excluded, Google said.
  • Google said Wiz found zero-days at two SaaS providers within two hours using Carbon for cyber.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

2 min read at securityweek.com

Google opened Carbon to paying developers and businesses as DeepMind used agents built on it to prove libsignal security guarantees. The post Google Widens Gemini 4 Carbon Access After AI Agents Verify Signal Code appeared first on SecurityWeek.

SecurityWeek

By Eduard Kovacs · securityweek.com

Read the full article at SecurityWeek

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.