🔥 Book Your Q1 Offensive Service Needs Now & Get a Discount! 🔥

Cloud Assumed Breach

One Stolen Login.
How Far?

Red Siege testers start from a compromised cloud identity and show how far an attacker could go in your Microsoft 365 and AWS environments.

Red Siege knight illustration: Cyber Knight

Assumed Breach for Microsoft 365 and AWS

Your email, files, and infrastructure now sit behind cloud logins and keys. If one is phished, leaked in a code repository, or exposed by a compromised server, what can an attacker reach? A Cloud Assumed Breach assessment answers that with simulated attacks against your Microsoft 365 and AWS environments, starting from a realistic compromised identity. We show where access policies, MFA coverage, and stored secrets give an attacker more than you intended, and how to better protect your data and assets.

Get Started

A Compromised Identity

Testing starts from a realistic cloud identity: an employee’s account, or the AWS role a compromised server uses.

Microsoft 365 and AWS

Simulated attacks across your Microsoft 365 and AWS environments, following the access an attacker would actually have.

Policies Tested, Not Assumed

We find conditional access and IAM permission gaps that grant more than you intended, before an attacker does.

Fixes You Can Verify

Findings rated by business risk, with short- and long-term recommendations and a validation step for each fix.

Our Approach

How We Test Your Cloud

Every engagement follows the same four steps, scoped to your Microsoft 365 and AWS environments. For AWS, our checks build on the CIS AWS Foundations Benchmark and AWS Foundational Security Best Practices.

Red Siege knight illustration: Elder Knight
  1. Start with what would hurt

    Before testing we ask which data or process in your cloud, if lost, stolen, compromised, or destroyed, would have the biggest impact on your organization. We also ask what your auditors require, because compliance needs like PCI can change what we test.

  2. Choose a realistic starting identity

    We agree where the breach begins: a user’s account, leaked or insider AWS access keys, a cloud server taken over through a web application flaw, or a developer’s access. Architecture diagrams, details on the managed services you use, and how we will connect let us spend the time testing, inside the cloud provider’s rules.

  3. Follow the access, not the label

    From that identity we test what it can actually reach: login portals that skip MFA, permissions broader than their names suggest, secrets stored in parameters, container images, function code, or storage snapshots, and development credentials that also work in production. We prove each escalation path actually works instead of reporting what a scanner flags.

  4. Report for leaders and engineers

    An executive summary that names root causes, technical findings rated by business risk, short- and long-term fixes, and a validation section for each finding so your team can confirm the fix without waiting for a retest. A missing control is never just a checkbox: each finding shows what an attacker could do because of it.

Test the Access You Actually Granted

In the cloud, an attacker with one valid identity may not need an exploit at all. A conditional access policy that misses one platform, or an AWS role meant to be read-only that can pull passwords and secrets, turns a single stolen login into a much larger breach. We test the access you have granted, not the access you think you granted,

because read-only doesn’t always mean read-only.

Get Offensive Now

“As red teamers, we’re supposed to show our clients what is versus what they assume is.”

Douglas Berdeaux, Senior Security Consultant, on why Red Siege tests AWS environments against what clients assume is in place. Watch the SiegeCast

Details on Cloud Assumed Breach

On a recent assessment, Red Siege found AWS roles with the ReadOnlyAccess policy that could read far more than the client intended. Mike Saunders shows how that access exposes EC2 administrator passwords, Parameter Store secrets, container images, and Lambda code, and how to build a safer replacement policy.

Learn More