Thousands of Credentials in Hours: How AI Agents Are Changing Cyberattacks in 2026

"The goal isn't to stop AI it is to control its permissions, monitor its actions and maintain meaningful human oversight"


Cybercriminals are moving beyond simple AI prompts. Here's what the rise of agentic AI means for cybersecurity and why speed may become the next major security problem.

 A Cyberattack That Changes the Question

Imagine a cyberattack that doesn't simply use AI to write a phishing email.

Instead, AI helps plan the operation, perform tasks, adapt to obstacles and continue working with much less human involvement.

That is the direction cybersecurity researchers are watching closely.

On September 8, 2026, Google Threat Intelligence Group reported that it observed threat actors using an autonomous, multi-agent AI framework in a ...Read More

The campaign was planned, built and executed in less than six hours and compromised thousands of third-party ...Read More

And that leads to a much bigger question:

What happens when cyberattacks become faster than humans can ...Read More

 From AI Assistants to AI Agents

There is an important difference between traditional AI assistance and agentic AI.

A normal AI chatbot might help an attacker write code or explain a technical problem.

An AI agent can potentially go further.

It can be designed to:

  • Plan tasks
  • Use tools
  • Interact with systems
  • Process information
  • Make decisions between steps
  • Continue working toward a goal

That extra capability is exactly what makes AI agents useful and potentially dangerous.

NIST has warned that AI agents create security challenges because AI models are being combined with software that can take actions in ...Read More

The risk isn't simply that AI can think.

The risk is what happens when AI can act.

 The Speed Problem

Cybersecurity has always been a race.

Attackers find vulnerabilities.

Defenders investigate.

Security teams detect suspicious activity.

Administrators block accounts and rotate credentials.

But humans need time to perform these tasks.

AI-driven automation can reduce the time required for many repetitive operations.

Google's latest threat research describes this transition as a move from basic prompting toward agentic AI workflows and AI-enabled automation, reducing the amount of human involvement needed during parts of ...Read More

That creates a new security challenge:

The attacker may not need to be faster than your security team.

The attacker may simply need to automate more of the work.

 Why Credentials Are Such a Valuable Target

Credentials are one of the most important pieces of the cybersecurity puzzle.

A stolen credential can potentially provide access to:

  • Cloud services
  • Email accounts
  • Developer environments
  • Databases
  • APIs
  • Internal applications
  • Source-code repositories

This is why credential theft remains such an attractive objective for attackers.

Google reported that the observed AI-assisted campaign compromised thousands of ...Read More

And there's another important lesson here:

AI doesn't need to invent a completely new type of cyberattack to create a major problem.

It can make existing attack techniques faster, more scalable and ...Read More

 AI Doesn't Have to Be Fully Autonomous to Be Dangerous

This point is extremely important.

When people hear about AI-powered cyberattacks, they sometimes imagine an AI completely replacing a ...Read More

That's not necessarily what is happening.

Google's research describes a gradual evolution.

Attackers are combining:

Human decision-making + AI models + automated tools + predefined instructions

to create increasingly capable workflows.

So instead of:

Human → Tool → Human → Tool → Human

the workflow can become closer to:

Human sets objective → AI agents perform multiple tasks → Human supervises

That difference can significantly reduce the time between ...Read More

 AI Agents Can Also Become Targets

There's another side to this problem.

AI agents aren't only being used by attackers.

They can also become targets.

Modern AI agents may interact with websites, emails, documents, APIs, code repositories and other external content.

That creates opportunities for prompt injection and agent hijacking.

NIST has specifically highlighted indirect prompt injection as a risk when malicious instructions are hidden inside content an AI ...Read More

OpenAI similarly describes prompt injection as an evolving security problem where third-party content can manipulate an AI system into doing something ...Read More

So imagine this simple situation:

User → AI Agent → Website

The user asks the agent to research something.

The agent visits a website.

Hidden inside that website is malicious content.

The agent reads it.

If the system isn't properly protected, that content may attempt to influence what the ...Read More

The website wasn't supposed to be giving the agent instructions.

But the agent may not always perfectly distinguish data from instructions.

That's one of the reasons agent security is becoming such an important topic.

 So How Do We Secure AI Agents?

The answer isn't simply:

"Don't use AI."

AI can provide enormous benefits.

The better question is:

How do we give AI useful capabilities without giving it ...Read More

Here are some important security principles.

1. Least Privilege

Give an AI agent only the permissions it actually needs.

If an agent only needs to read a particular folder, why should it have access to the entire organization's ...Read More

2. Human Approval for High-Risk Actions

Some actions should require human confirmation.

For example:

Reading information may be low risk.

Sending money, deleting data or changing critical settings can be very different.

The higher the potential impact, the stronger the human ...Read More

3. Monitor Agent Activity

Security teams need visibility into what agents are doing.

Logging should help answer questions such as:

  • What did the agent access?
  • Which tools did it use?
  • Which files did it read?
  • What actions did it perform?
  • What permissions did it use?

Google's cybersecurity forecast has emphasized the need for distinct authorization, authentication and monitoring ...Read More

4. Treat External Content as Untrusted

A webpage, email, document or code repository may contain useful information.

But it should not automatically be treated as a trusted instruction source.

This is especially important for defending against ...Read More

5. Have a Stop Mechanism

What happens when an AI agent starts behaving unexpectedly?

There should be a way to:

Pause → Investigate → Revoke access → Shut down

if necessary.

This isn't about assuming AI will always behave badly.

It's about preparing for the possibility that ...Read More

 AI Is Also Becoming Part of the Defense

Here's the interesting part.

The same technology attackers are using can also help defenders.

Security teams can use AI to help:

  • Analyze alerts
  • Investigate logs
  • Find suspicious behavior
  • Identify vulnerabilities
  • Prioritize security issues
  • Automate repetitive defensive tasks

Google's 2026 cybersecurity forecast describes this emerging model as an "Agentic SOC", where AI agents can assist ...Read More

So cybersecurity may increasingly become:

AI vs AI with humans responsible for the important decisions.

 The Bigger Cybersecurity Lesson

The biggest lesson isn't that:

"AI is evil."

And it isn't that:

"AI will replace hackers."

The more practical lesson is this:

Every new capability creates a new security responsibility.

When software could only store information, we worried about unauthorized access.

When software became connected to the internet, we worried about network attacks.

When cloud computing became widespread, identity and access management became ...Read More

Now AI systems are becoming capable of making decisions and taking actions.

That means cybersecurity needs to ask a new question:

Who or what is allowed to act?

 The Future of Cybersecurity May Depend on Identity

Organizations already manage identities for:

  • Employees
  • Administrators
  • Applications
  • Services
  • Devices

AI agents are creating another category.

Machine identities that can make decisions and perform actions.

Google has argued that organizations will need to treat AI agents as distinct digital actors with ...Read More

That means the future security architecture may need to answer:

Which agent is this?

What is it allowed to access?

Who authorized it?

What action is it performing?

Why is it performing that action?

Can we stop it?

Those questions may become just as important as traditional username-and-...Read More

 Final Thought

AI-powered cyberattacks don't have to look like science fiction.

They can begin with something much simpler:

A stolen credential.

A compromised cloud account.

An automated workflow.

An AI agent given too much access.

And suddenly, an attack that once required many manual ...Read More

The lesson for cybersecurity professionals is clear:

Don't only secure the AI model. Secure what the AI is allowed to do.

Because intelligence without boundaries can become a security problem.

And in the age of agentic AI, permissions may matter just as...Read More

Comments

  1. Tahami Rehman BlogsSeptember 9, 2026 at 8:18 AM

    What worries me most isn't simply that AI can become more capable—it's how much access we give it.

    ReplyDelete

Post a Comment

Popular posts from this blog

AI Agents Are Getting Powerful But Who Is Controlling Them? The Cybersecurity Risk We Must Understand in 2026

Image
"AI agent security illustration showing prompt injection, email and web content, APIs, least privilege, human approval, monitoring, audit logs and zero-trust controls" Imagine giving an AI assistant access to your email, files, cloud account, APIs and business tools. You ask it to do a simple task. It reads an email. Inside that email is hidden text designed to manipulate the AI. The AI follows the hidden instruction instead of your ... Read More Now the question becomes: Did the attacker hack the AI or did the AI become the path to your systems? This is one of the important security questions emerging as AI agents become more capable.  What is an AI agent? A normal chatbot mainly responds to your questions. An AI agent can go further . Depending on how it is designed, an agent may: Read emails Search websites Work with files Call APIs Use business applications Write or execute code Take actions on behalf of a user That extra ability creates a ...
Read more

AI Agents Are Moving From Chat to Action What Does This Mean for Cybersecurity in 2026?

Image
"Smarter AI needs stronger security. Capability is powerful, but responsible access matters even more" For years, we mostly used AI to ask questions and receive answers . Now something bigger is happening. AI agents are increasingly being designed to take actions . They can work with files, use APIs, interact with websites, search information, write code ... Read More That sounds exciting. But it also creates a new cybersecurity question: What happens when an AI system has the ability to act but something causes it to act in the wrong way? This is becoming one of the important conversations ... Read More  From AI Assistant to AI Agent A traditional AI chatbot might answer: "How can I organize my emails?" An AI agent could potentially go further: Read emails → identify important messages → organize them → update another application → report the result. The difference is simple: Chatbot = mainly responds Agent = can reason, use tools and take ac...
Read more

AI Agents Are Changing Cybersecurity in 2026 — 7 Security Risks You Should Know

Image
 "AI systems gain the ability to use tools and take actions, protecting their permissions, data, credentials and access becomes a critical part of cybersecurity" AI is no longer just answering questions. It can now plan, use tools, and take actions and that changes cybersecurity. Artificial intelligence has moved far beyond simple chatbots. Modern AI systems can analyze information, write code, interact with applications, use tools, and perform multi-step tasks. This new generation is often called AI agents  ... Read More That creates a new cybersecurity question: What happens when an AI system has access to sensitive data, accounts, applications, and tools? The answer is simple: security becomes even more important. Google Threat Intelligence reported in September 2026 that some threat actors are moving toward agentic AI workflows that can automate parts of cyber operations and reduce the amount of ... Read More So let's understand the risks in simple langu...
Read more