Vulnerabilities & attacks · Briefing

Cyberattack on Quantum Systems:When the update channel itself becomes a risk

Media reports describe a malicious file distributed as an update for the QBase ground-control and mission-planning software. The Bundeswehr reportedly grounded drones as a precaution. The manufacturer confirms the attack and containment, but not the technical details. Our briefing separates the evidence and shows why the CRA, threat modeling and a controlled product lifecycle belong together.

  • Detected in late October 2026
  • QBase update path according to reports
  • No drone infection established
  • Cause and actors unresolved

At a glance

What is established – and what is not

The reports concern not primarily the drone, but the trust relationship between manufacturer and customer. An update connects engineering, build infrastructure, customer networks and operations. If this chain is uncertain, a precautionary operational pause may be necessary even without a demonstrated endpoint compromise.

  1. The manufacturer confirms the attack. Quantum Systems states that it detected the incident in late October, contained it with authorities and reported it.
  2. The update path comes from media reports. The public manufacturer statement does not name QBase, the update server or distribution to customers.
  3. No infected drone is established. The Der Spiegel report describes detection during a check before the file reached aircraft.
  4. The grounding was precautionary. Reports differ on scope and duration; the Ministry of Defence has not publicly confirmed the details.
  5. The CRA is relevant but not universally applicable. Reporting duties have applied since September 2026; products developed or modified exclusively for defence are excluded.
  6. Secure updates require a lifecycle. Threat modeling, SAST, SCA, DAST, fuzzing, SBOM, release gates and PSIRT must work together.

Confirmed or consistently reported

  • Quantum Systems confirms an active cyberattack by hostile actors and cooperation with authorities.
  • The Financial Times and Der Spiegel report a malicious file disguised as a QBase update.
  • The file reportedly reached customer servers and was detected during a check, not on the drones themselves.
  • On November 13, the manufacturer stated that the fleet, including the Bundeswehr, was operating normally again.

Still unresolved

  • How attackers entered the manufacturer or update infrastructure
  • What the file was intended to do and which versions were affected
  • Whether a signing or release process was compromised
  • Who carried out the attack and which customers received the file
  • Whether CRA reporting was required and which reporting path was used
Timeline

From detected incident to public briefing

The timeline separates manufacturer statements from media reports. It does not assume a specific detection or grounding date.

  1. Incident

    Quantum Systems detects the incident

    The company states that it identified a cybersecurity incident in late October and determined that it was an active attack. [1]

  2. Manufacturer

    Media reports and manufacturer statement

    Media report on QBase, the update server and precautionary grounding. Quantum Systems confirms the attack without confirming these technical details. [1][2]

  3. Manufacturer

    Manufacturer reports normal flight operations

    A spokesperson states that the fleet is operating normally worldwide, including Germany and the Bundeswehr. No independent confirmation is available. [3][4]

  4. Analysis

    Technical and regulatory questions remain

    The actors, malware function and entry vector remain unclear. The public information does not allow a concrete CRA assessment. [1][2][5]

Attack path

Why the update channel is an attack surface in its own right

This chain reflects the reported situation. It does not show which specific control failed; that is not publicly known.

  1. Engineering

    Code, dependencies and build configuration form the origin of a release. Vulnerabilities or compromised accounts can enter here.

  2. Build and signing

    The reviewed source becomes an artifact. Provenance, a protected pipeline and signing must establish origin and integrity.

  3. Distribution

    The update server is a trust boundary between the manufacturer and all customers. A manipulated file can spread through the same mechanism as legitimate updates.

  4. Customer-side verification

    Signature and integrity verification, staging and controlled installation must prevent unexpected packages from taking effect in production.

  5. Operations and release

    Monitoring, staged rollout, revocation and a precautionary pause limit impact while the situation is unclear.

What is not publicly established

There is no public evidence of infected drones, a compromised code-signing key, a particular threat actor or successful execution on mission computers. None of these points should be inferred from the reports.

Why the impact extends beyond IT

  • Update infrastructure connects one manufacturer system with many customers and operating environments.
  • Ground software can be security-relevant for mission planning, flight monitoring and data processing even when it does not run on the aircraft.
  • Unclear integrity forces operators to make risk-based decisions, including a grounding.
  • A robust release process must support stopping, revocation and safe recovery as well as delivery.
Cyber Resilience Act

CRA: update security becomes a manufacturer process

For products with digital elements, the CRA requires risk-based development, regular security testing, SBOM-supported vulnerability management, coordinated reporting channels and mechanisms for secure update distribution. The BSI explicitly names a manipulated update server as an example of a severe incident affecting product security.

  1. In force

    The CRA has entered into force; obligations apply in stages.

  2. Assessment bodies

    Rules for notifying conformity assessment bodies apply.

  3. Reporting duties

    Early warning within 24 hours and detailed notification within 72 hours; the final report follows after 14 days or one month, depending on the case.

  4. Full application

    The essential product and lifecycle requirements apply in full.

Risk-based development

Product risk assessment must cover planning, design, development, production, delivery and maintenance. Update servers, build systems and signing services belong in this picture where they influence the product.

SBOM and vulnerabilities

Components and vulnerabilities must be documented, tested regularly and remediated without delay according to risk. A machine-readable SBOM belongs in the technical documentation.

Secure update distribution

Annex I Part II requires secure distribution mechanisms. Signature verification, protected keys, provenance and controlled delivery are natural technical implementations, not a rigid single prescription.

Manage reporting deadlines

The deadline starts when the manufacturer becomes aware. Prepared decision paths, product context and a complete incident timeline are operational prerequisites.

Support and disclosure

Vulnerabilities must be handled effectively and security updates provided during the support period, including coordinated disclosure and user information.

Evidence instead of assertions

Technical documentation must show how requirements are implemented. Release gates, test results, SBOM, approvals, incident history and customer communications form the auditable connection.

Important: no blanket CRA application to this defence case

Article 2(7) of the CRA excludes products developed or modified exclusively for national security or defence purposes. Public information is insufficient to determine which CRA duties apply to QBase or a specific system. This briefing uses the CRA as a reference framework without alleging an infringement or a specific reporting obligation. [7][8]

Secure Product Lifecycle

From architecture risk to a controlled update

Product security is not a single scan before release. Each phase needs a concrete risk question, effective controls and evidence connected to the shipped release.

  1. Plan and model

    Architecture, data flows, trust boundaries

    Typical risk
    Update server, signing service or build system is missing from the risk picture.
    Controls & evidence
    STRIDE model, product risk assessment, scope decision and accountable owners.
  2. Engineer

    Code, secrets and dependencies

    Typical risk
    Proprietary code or third-party components introduce vulnerable logic into the build.
    Controls & evidence
    SAST, SCA, secrets scanning, review approvals and findings with owners.
  3. Build and test

    CI/CD, artifact and runtime

    Typical risk
    The build or update package differs from the reviewed state; runtime flaws remain invisible.
    Controls & evidence
    SBOM, build provenance, signing, DAST and fuzzing, including negative update tests.
  4. Release and distribute

    Release gate, staged rollout, revocation

    Typical risk
    A faulty or manipulated release reaches every customer without a control point.
    Controls & evidence
    Signed approval, integrity verification, rollout log, revocation and rollback test.
  5. Operate and report

    SBOM watch, PSIRT, customer communication

    Typical risk
    Exposure, priority and reporting deadlines remain unclear during an incident.
    Controls & evidence
    Release inventory, VEX/VDR, incident timeline, reporting drafts and documented decision.

Six controls for every update channel

  1. Signing: Verify authenticity and integrity before installation; protect keys and privileges.
  2. Provenance: Connect source revision, build process and artifact in a traceable way.
  3. SBOM: Document and monitor direct and transitive components for each release.
  4. Negative tests: Test invalid signatures, invalid metadata, downgrades and interrupted installation.
  5. Stages and revocation: Control, stop and withdraw rollout and recover safely.
  6. Monitoring and PSIRT: Know shipped versions, escalate incidents, inform customers and manage deadlines.

The release as an evidence package

A robust process connects threat model, risk decision, SBOM, scan and test results, build provenance, release approval and incident history. It demonstrates which controls actually worked instead of merely asserting that an update was secure.

Agentic threat modeling

STRIDE applied to the update path

These examples are our methodological assessment. They do not describe facts about this attack; they show how VamiThreat structures threats and controls across components and trust boundaries.

S · Spoofing

Forged release identity

Misused build or service accounts and impersonated services require strong workload identities, MFA and protected keys.

T · Tampering

Manipulated update package

Code, build, metadata and package must be protected through signing, provenance and immutable artifacts.

R · Repudiation

Unclear approval

Audit logs, named release owners and a traceable rollout history prevent unclear decision paths.

I · Information Disclosure

Secrets and customer data

Secrets scanning, least privilege and encryption limit the exposure of credentials and sensitive information.

D · Denial of Service

Operations interrupted

Staged rollout, tested revocation, rollback and recovery reduce impact on the update service and customer operations.

E · Elevation of Privilege

Account gains release rights

Least privilege, separation of duties and just-in-time access limit what a compromised account can become.

From model to action

VamiThreat turns architecture information into a reviewable data-flow model, enumerates STRIDE threats and adds risk, attack path and concrete mitigation. MAESTRO is available for agentic AI systems. Results can feed risk and remediation management.

Self-check

Could you stop a manipulated update today?

Six questions for product and security leaders. “Yes” is always the safe answer. The assessment runs only in your browser and does not replace a technical review.

  1. Are build, signing and update systems included in the current threat model?
  2. Can every shipped release be assigned an SBOM and verifiable build provenance?
  3. Are packages with invalid signatures or metadata reliably rejected?
  4. Do SAST, SCA, DAST and fuzzing examine code, dependencies and runtime behavior before release?
  5. Are staged rollout, stopping, revocation and safe recovery tested?
  6. Are CRA scope, PSIRT owners and reporting deadlines clarified before an incident?
Your assessment

Answered: 0 of 6 questions. The assessment appears after the final answer.

Your answers are neither stored nor transmitted.

Action plan

What product teams should do now

These measures are general hardening steps for an update and product supply chain, independent of this specific incident.

Immediately

Make the update path visible

  • Record build, signing and update infrastructure as critical assets.
  • Inventory current releases, components, signatures and distribution paths.
  • Review pipeline privileges, service accounts and external dependencies.
  • Clarify revocation, rollback and customer contact for an unsafe release state.
30 days

Test controls and collect evidence

  • Integrate SAST, SCA, secrets scanning, DAST and fuzzing into appropriate pipeline gates.
  • Generate SBOM and build provenance for each release and document signed approvals.
  • Demonstrate negative update tests, staged rollout and revocation.
  • Test PSIRT, CRA scope and the reporting decision in a tabletop exercise.
90 days

Embed the lifecycle

  • Update the threat model when architecture or releases change.
  • Continuously match SBOMs of shipped releases against new risk signals.
  • Connect risks, actions and evidence in a product security management system.
  • Make supplier and customer requirements for secure updates contractually and technically verifiable.
Suspect a manipulated release process?

Stop distribution, identify affected versions and customer contacts, preserve evidence and timestamps, control access to signing and release systems and assess reporting obligations case by case. Our CSIRT supports you around the clock.

Report an incident csirt@vamisec.com
How VamiSec supports you

From threat model to controlled release

VamiThreat models threats, VamiAppSec tests code, dependencies and runtime, and VamiGRC manages risks, actions, evidence and reporting. The three solutions connect along the same product lifecycle.

VamiGRCAgentic CSMS & GRC

VamiGRC: manage risks and evidence across the product lifecycle

VamiGRC connects product, release, supplier, risk, control and incident in the GRC Graph. Agentic workflows support risk assessments, release gates, remediation planning, evidence collection and reporting drafts – with defined approvals for your team.

VamiGRC is an agentic integrated management system for manufacturers and operators: NIS2 for your organization and the CRA for your products in one system. ISMS, PIMS, AIMS, CSMS, and BCMS share the same assets, suppliers, risks, and evidence on one GRC graph; risks and measures are assessed in euros. AI agents handle the routine work according to your approval rules.

Free until go-live: one month of Proof of Value with demo or real data, followed by guided implementation after a binding order – the license starts only at go-live.

For the update and product supply chain
  • Product Security Lifecycle with release gates and accountable owners.
  • SBOM, KEV and NVD context for each release, connected to products and assets.
  • Risks, actions and evidence on shared data instead of separate registers.
  • CRA and NIS2 deadlines in the incident process; VamiAI drafts reports, your team reviews and submits them.
  • Autonomy levels L0 to L3 define what runs automatically and what requires approval.

VamiAppSec

Secure SDLC & product security

VamiAppSec orchestrates SAST, SCA, SBOM, secrets, IaC and container scans as well as AI-powered DAST in your CI/CD. Fuzzing complements testing of APIs and update clients. Findings are tracked and become usable release evidence.

  • SAST for proprietary code and SCA for direct and transitive dependencies.
  • SBOM per release and continuous matching against exploitation signals.
  • AI-powered DAST and API contract fuzzing for runtime behavior.
  • Quality gates, ticket tracking and traceable remediation through to release evidence.

VamiThreat

Agentic threat modeling

VamiThreat turns architecture information into a reviewable data-flow model with trust boundaries and enumerates threats using STRIDE. MAESTRO is available for agentic AI. Each threat receives risk, attack path and remediation context.

  • STRIDE across components, data flows and trust boundaries.
  • Update, build, release and operational risks in one model.
  • MITRE ATT&CK, ICS and ATLAS mapping per component.
  • Prioritized developer-ready actions instead of a generic risk list.
Valeri Milke, founder and managing director of VamiSec GmbH
Schedule a call

Let’s talk about your update supply chain

Whether threat modeling, CI/CD hardening, DAST and fuzzing, CRA roadmap or PSIRT process: Valeri Milke and the VamiSec team help you derive concrete next steps from this briefing.

“IT security and compliance without AI is a vulnerability.”

Valeri MilkeFounder & Managing Director · VamiSec GmbHISO/IEC 27001 & ISO/IEC 42001 Lead Auditor
20Years of GRC experience of the founder
ISO 27001VamiSec is itself certified
T Cloud PublicOperation in German data centers
24/7CSIRT for security incidents
FAQ

Frequently asked questions about the QBase update incident and the CRA

Were the drones themselves infected with malware?

The reviewed reports do not establish this. The Der Spiegel report describes the file reaching customer servers and being detected during a check, not an infection of the aircraft. Installation on mission computers has not been publicly confirmed either.

Has Quantum Systems confirmed the technical details?

The public statement confirms the cyberattack, detection in late October, containment and reporting to authorities. It does not name QBase, the update server, malware function or entry vector. Those details come from media reports.

Is Quantum Systems required to report under the CRA?

This cannot be determined conclusively from public information. The manufacturer states generally that it reported the incident. CRA reporting requires a review of product scope, severity and impact; products developed or modified exclusively for defence or national security are excluded.

Does the CRA require signed updates?

Annex I Part II of the CRA requires mechanisms for secure update distribution, but does not prescribe one specific technology. Signature verification, protected keys, build provenance and secure delivery are natural implementations; selection and risk justification remain product-specific.

What do SAST, SCA, DAST and fuzzing each contribute?

SAST examines code without execution, SCA identifies dependency risks, DAST tests the running system from outside, and fuzzing tests robustness and error behavior with varied or unexpected input. None replaces the others; together with threat modeling, SBOM and release controls they provide broader protection.

How do VamiGRC, VamiAppSec and VamiThreat support this?

VamiThreat supports STRIDE and MAESTRO models, VamiAppSec orchestrates AppSec tests and release evidence, and VamiGRC connects product lifecycle, risks, actions, suppliers and reporting. The responsible team retains expert decisions and required approvals.

Sources

Sources and further information

The information is based on the manufacturer statement, media reports and official CRA information reviewed through November 14, 2026. Reported technical details and open questions are identified as such.

  1. [1]Quantum Systems: Statement on Cyber Incident,
  2. [2]Dealroom: A hacked software update grounded Quantum Systems’ drones in the German military,
  3. [3]AeroTime: Quantum Systems says Bundeswehr drones flying again after cyberattack,
  4. [4]The Register: Quantum Systems drones back in the air after malware scare,
  5. [5]DroneXL: Quantum Systems Hack: Fake Update Got German Military Drones Grounded,
  6. [6]Europäische Kommission: Cyber Resilience Act – Reporting obligations,
  7. [7]BSI: Cyber Resilience Act: Anforderungen, Scope und Meldungen,
  8. [8]EUR-Lex: Verordnung (EU) 2024/2847 – Cyber Resilience Act,

This briefing is no substitute for legal advice. Which obligations apply to a product or company depend on the individual case.

Act now

Updates require trust.
We help build the controls and evidence.

Arrange a no-obligation initial consultation: we will clarify your starting point and the next steps for a secure product lifecycle.