Google bets $32B on AI agent cyber force as security arms race escalates

Viktor Tanasiichuk via iStock / Getty Images Plus

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• Google launches AI agents for cyber defence warfare.
• $32bn Wiz woody signals nation-state level urgency.
• AI present hunts, detects, and fixes threats astatine speed.

Today astatine Google Cloud Next 2026, Google is introducing a caller agentic defence portfolio that combines menace intelligence, information operations, and proactive menace mitigation activities.

Also: 5 information tactics your concern can't get incorrect successful the property of AI - and wherefore they're critical

In different words, Google is going to war, and it's unveiling its large guns.

Humans aren't accelerated capable

Cyberattacks person been portion of the computing scenery since determination person been networked computers. As the web grew larger and faster, the ferocity of attacks increased.

Whether initiated by nation-states, criminals, hacktivists, oregon disgruntled individuals, attacks person ever been asymmetric. In different words, each the attacker has to bash is find 1 flaw to usage arsenic an introduction point. Defenders person ever had to support against everything.

Also: AI agents are fast, loose, and retired of control, MIT survey finds

While assaults similar denial-of-service attacks could tally astatine instrumentality speed, the capableness to make and deploy attacks was ever constricted by humans' quality to find exploits and plan attacks. Likewise, the defenders could initiate automated defenses, specified arsenic firewalls. However, onslaught mitigation had to beryllium done by radical with discernment truthful they could adhd protections and respond without breaking systems.

AI changes each of that. Enemy actors tin usage enormously almighty ample connection models to place vulnerabilities and deploy attacks astatine electron speed. By utilizing parallel agents, they tin adjacent bash truthful with tremendous integer armies of attackers, each moving astatine speeds good beyond the powers and abilities of mortal humans.

Also: Will AI marque cybersecurity obsolete, oregon is Silicon Valley confabulating again?

To support against larger onslaught surfaces, faster AI deployment, and adversaries utilizing AI for much blase attacks, the bully guys besides request AI armies. Human analysts can't process the barrage of bits accelerated enough.

That operation of alteration brings america to Google; the steadfast is fundamentally launching a cyberforce of AI agents that tin not lone run connected the beforehand lines of cyberwar but besides proviso back-end logistics and quality analysis. That attack is astatine the halfway of this announcement.

$32 cardinal

Wiz is simply a cybersecurity institution formed successful 2020. Its assertion to fame is an uncanny quality to find faults and vulnerabilities successful networks and bundle platforms. Since its founding, Wiz has efficaciously go the apex predator of cybersecurity.

Just past month, Google's genitor Alphabet acquired Wiz. All it took was a $32 cardinal all-cash transaction, the largest ever cybersecurity acquisition, and the azygous biggest acquisition successful Alphabet history.

Also: Why endeavor AI agents could go the eventual insider threat

Wiz, according to Alphabet, "Delivers an easy-to-use information level that connects to each large clouds and codification environments to assistance forestall cybersecurity incidents."

Let's deliberation astir $32 billion, a fig that's much than Canada's full military defence budget and astir arsenic overmuch arsenic Israel's subject spending. Laying retired $32 cardinal connected a cybersecurity acquisition tells america 2 things: the menace is real, and it justifies nation-state-level spending by the tech giant.

Threat quality

Ancient Chinese subject general, strategist, and philosopher Sun Tzu said: "If you cognize the force and cognize yourself, you request not fearfulness the effect of a 100 battles. If you cognize yourself but not the enemy, for each triumph gained, you volition besides endure a defeat. If you cognize neither the force nor yourself, you volition succumb successful each battle."

This conception is astatine the halfway of menace intelligence. In a cybersecurity context, knowing yourself means knowing your vulnerabilities and being capable to way and negociate force attacks and invasions. If an attacker tin sneak into your web and unrecorded determination for weeks oregon months, you don't cognize yourself.

Also: AI threats volition get worse: 6 ways to lucifer the tenacity of your integer adversaries

Google is announcing Agentic SecOps (security operations) with 3 cardinal prongs.

The tech elephantine uses Gemini AI to research the acheronian web and physique "a nuanced illustration of your organization." The AI tin "analyze millions of regular outer events with 98% accuracy to assistance elevate lone the threats that genuinely substance to your organization."

Google is besides deploying a caller threat-hunting cause that uses the immense menace quality cognition gathered crossed its infrastructure to "proactively hunt for caller onslaught patterns and adversary behaviors that bypass accepted defenses."

In addition, Google is deploying a detection engineering agent. This beastie automatically generates persistent menace detection rules. The attack is similar having a robot constitute super-smart firewall rules automatically, but for each levels of web threats.

Also: 10 ways AI tin inflict unprecedented harm successful 2026

Because the atrocious guys person entree to AIs that tin rapidly deploy caller threats, defenders besides request to beryllium capable to leap the quality velocity obstruction and deploy caller antiaircraft engineering solutions astatine instrumentality speed.

According to Google, "Customers are already benefiting from our Triage and Investigation Agent, which has processed much than 5 cardinal alerts to date, reducing a emblematic 30-minute manual investigation to 60 seconds."

Red, Blue, and Green

The Wiz constituent plays a relation by protecting AI and unreality apps crossed immoderate infrastructure. For immoderate broad antiaircraft solution to beryllium effective, it has to beryllium disposable crossed vendor merchandise lines.

The Wiz AI Application Protection Platform supports Databricks, AWS Agentcore, Gemini Enterprise Agent Builder, Microsoft Azure Copilot Studio, and Salesforce Agentforce. Wiz besides offers cloud-edge protection, extending its shields astir implementations from Apigee, Cloudflare, Vercel, and "others."

Also: Why encrypted backups whitethorn neglect successful an AI-driven ransomware era

A large payment is that multivendor enactment besides adds much discourse astir the outer onslaught surface, meaning the exertion understands the menace situation much completely.

For active-threat situation defense, Wiz is deploying Red, Green, and Blue Agents that enactment arsenic a information quality squad crossed the enterprise.

The Red Agent is simply a penetration investigating information researcher. It's designed to find ways into your web and past catalog that accusation for the different agents successful the network. Think of the Red Agent arsenic a information defender perpetually patrolling and trying each the locks to marque definite they're really inactive locked.

Then deliberation of the Blue Agent arsenic a transgression country detective. It gathers grounds from logs, identities, and strategy activity, and uses that accusation to reconstruct behaviors and find severity. Its occupation is to enactment arsenic a forensic expert who discovers each the details of a breach and explains the communicative down what happened.

Also: 5 ways you tin halt investigating AI and commencement scaling it responsibly successful 2026

The Green Agent is the maestro mechanic. Given accusation from the Red and Blue Agents, the Green Agent goes retired and builds a fix. Key to the AI show is that it builds a focused fix, specifically tied to the existent network. That way, a hole has a overmuch little accidental of undoing thing already moving decently connected the network.

Together, the Red Agent looks for anemic points, the Blue Agent identifies however and wherefore thing atrocious mightiness person happened, and the Green Agent stops atrocious worldly from happening again. Think of this attack arsenic test, investigate, and fix.

ReCAPTCHA is truthful 2024

As acold backmost arsenic 2024, AIs could lick reCAPTCHA tests. You cognize those tests? They're designed to corroborate that you're a quality and not a bot trying to spoof thing connected the internet.

ReCAPTCHA works, to a point. We've each been frustrated that the 5th representation is simply a span oregon a motorcycle, and for immoderate reason, reCAPTCHA doesn't admit it arsenic such. Raise your manus if you've yelled "I'm human" astatine your machine much than once. I have. I'm not proud.

Also: 5 ways to usage AI to modernize your bequest systems

Into this space, Google is introducing Google Cloud Fraud Defense. Google describes this arsenic "The improvement of reCAPTCHA, and provides the quality that businesses request to spot their integer interactions and commerce."

It's fundamentally a level designed to find whether an accessing entity is simply a human, a bot, oregon an agent.

State the facts, Jack

Google included immoderate societal impervious successful its announcement. The institution described occurrence stories from a assortment of large customers utilizing these caller tools.

A fewer of those firms that person seen show improvements include:

• Colgate-Palmolive decreased outer vulnerability issues by 44% and present sustains agelong periods of zero captious risks with Wiz.
• Deloitte accrued expert ratio by much than 60%. Threat hunts crossed billions of logs dropped from hours to seconds. Detection regularisation procreation present takes minutes alternatively of weeks.
• Urgent and caller information vulnerabilities that utilized to instrumentality Shell betwixt 3 days and 2 weeks to observe are present managed successful near-real time.

Welcome to the 2026 arms contention

When multi-billion-dollar companies commencement spending connected defence similar nation-states and deploy AI agents similar battalions, it's clip to judge that the crippled has changed.

Attackers are scaling, automating, accelerating, and adding quality that thinks astatine warp velocity earlier quality defenders tin down their archetypal cupful of coffee. Malicious AIs tin tally 24 hours a day, 7 days a week, without needing slumber oregon caffeine. All they request to bash is find 1 error, and they're in.

To defend, targets request to run astatine superhero speed, prolong that attack astir the clock, and drawback and mitigate attacks faster than a blinking eye, faster than a speeding bullet, and faster than the clip it takes to click a mouse.

Google is surely not the lone large institution moving connected this problem, but they present person a viable introduction into the arms race. Unfortunately, an arms race, by definition, ne'er truly ends. It lone escalates.

How comfy are you with an AI strategy that builds and deploys its ain detection rules crossed your network? Let america cognize successful the comments below.

You tin travel my day-to-day task updates connected societal media. Be definite to subscribe to my play update newsletter, and travel maine connected Twitter/X astatine @DavidGewirtz, connected Facebook astatine Facebook.com/DavidGewirtz, connected Instagram astatine Instagram.com/DavidGewirtz, connected Bluesky astatine @DavidGewirtz.com, and connected YouTube astatine YouTube.com/DavidGewirtzTV.