Prompt Pack
Prompt Pack

Skip-Level

Google's Gemini 4 Carbon Tops an AI Index and Targets Cyber Defense

By ยท 29 Oct 2026

An IT manager, seen from behind at a standing desk, presses a teal tile into the last open slot of a wall grid of blue and coral tiles while small status lights glow along the top.
  • Google says its new Gemini 4 Carbon model leads on finding software vulnerabilities, with access limited to cyber defenders for now.
  • GitHub says agents now start more than 40% of pull requests on its platform.
  • Gartner finds just 4% of CIOs have plans to retool roles for a workforce of people and AI agents.

Three things engineering and IT managers should know this Thursday, from a security focused Google model most teams cannot use yet to new GitHub data on how much code agents now start, and a Gartner survey on who is redesigning roles around them.

1. Google's Gemini 4 Carbon Leads on Finding Vulnerabilities

Google announced Gemini 4 Carbon on Oct 28 and calls it its strongest model, with state of the art cybersecurity and long horizon capabilities. Google says Carbon ranks first on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59. On security benchmarks evaluated without guardrails, it reports 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. Google says its own teams have seen Carbon run for days on complex engineering tasks, such as root causing and patching subtle bugs across large codebases. Starting this week, Google is making Carbon available to a subset of cyber defenders and will gradually extend access to more trusted partners through its Fairwind Program, before it reaches developers, businesses, and consumers. The introductory price matches Gemini 4 Argon at $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after the introductory period.

Why it matters: Most teams cannot use Carbon yet, so there is nothing to pilot this week. Our read: the useful move now is to pick one real backlog, such as open findings from your last security scan, that you would test a model like this on once access opens, and to decide who reviews the patches it writes. Vendor benchmark scores are a starting point, not proof the model works on your code. Source: Google

2. Agents Now Start More Than 40% of GitHub Pull Requests

GitHub's 2026 Octoverse report says developers pushed code 17.9 billion times in the past 12 months, more than in 2024 and 2025 combined. As of September, more than 40% of pull requests on GitHub started with an agent's first draft, up from 17% in March, which GitHub counts as 110 million in September alone. Among repositories created this year, the average one with an agent configuration file logged eight times as many commits as the average new repository without one. Agent started pull requests also more often change multiple files, about 87% against 79%, and agent work lands: by September, 90% of agent involved pull requests merged within two weeks, against about 86% for pull requests without detected agent involvement.

Why it matters: GitHub itself says review is now how developers shape what agents send back. Our read: the eight times commit gap shows where agent users are, not that agents cause it. If a growing share of your team's pull requests start with an agent, review capacity and clear merge ownership matter more than raw output. Track how long changes wait for review, not how many pull requests your team opens. Source: GitHub Octoverse

3. Few CIOs Are Redesigning Roles for Agent Work

Gartner surveyed 2,002 CIOs and technology executives for its 2027 CIO and Technology Executive Agenda report, CIO Dive reported on Oct 19. Just 4% of CIOs said they have plans in place to revamp roles with a human and agent workforce in mind. Enterprises that do are 2.4 times more likely to be top performers with AI and to have their CEOs report value from AI. Meanwhile, 58% of leaders report increased demand from their companies to deliver cost improvements from prior AI investments, and just 13% of CIOs report significant or transformational value from their AI investments. Gartner VP analyst Alicia Mullery told CIO Dive that cost pressure can overshadow talent work because it is not as immediate as an expensive bill.

Why it matters: The role question lands on engineering managers before it reaches the org chart. Our read: the 2.4 times figure is an association, not proof that redesign causes AI success. Still, as agents take on more first drafts, the work on your team shifts toward review and direction whether or not anyone writes it down. Name this week who reviews agent work, who owns agent setup, and how you will measure output. Source: CIO Dive

Compiled daily with Grok from public sources and edited by the AIDevPulse team.

Get Skip-Level and the daily brief in your inbox: sign up for the AIDevPulse newsletter. For today's developer news, read today's main brief.

Sources:

Recommended AI prompt

Copy this paragraph into ChatGPT, Claude, Gemini, Grok, or whatever AI you use.

Prompt

I manage an engineering or IT team. Using only the three items in this readout, help me draft a one page plan for the next month that does three things: lists which kinds of pull requests we let agents start and who reviews and merges them, names one security backlog we would test a security focused model on once we have access and how we would judge its patches, and states which roles on my team change as agents take on more first drafts. Keep my team's context in mind: [describe your team's size, the AI tools you use today, and where work most often waits for review]. Do not add facts that are not in the readout, and flag anything I should check before sharing.