We own 100% of the FDA Section 524B premarket cybersecurity submission for your SaMD: threat model, secure architecture, SBOM, cloud + mobile + API pen testing, vulnerability management plan, labeling, and reviewer-ready evidence binder. Most modern SaMD includes AI/ML, so we extend the same package with PCCP, GMLP, adversarial robustness, and transparency labeling when your device uses ML. Postmarket monitoring is a separate annual contract, but you can bundle it with premarket at signature for a single fixed fee from kickoff through your first year on market.
The short answer
Software as a Medical Device (SaMD) that connects to the internet counts as a cyber device under Section 524B, so it needs the same deliverables as hardware: a threat model, risk assessment, SBOM, architecture views, penetration testing and a postmarket management plan. For SaMD, the cloud, APIs, mobile apps and identity systems are usually the main attack surface, so testing and documentation focus there.
One team owns the full FDA Section 524B premarket package, with AI/ML extensions (PCCP, GMLP, AAMI CR34971) layered in when your SaMD includes a model.
Trusted by leading MedTech companies
Most cybersecurity guidance assumes a connected device with firmware. When the device is a cloud service or a mobile app, the deliverables change.
FDA reviewers expect a cloud reference architecture (VPC, IAM boundaries, KMS, tenancy isolation, logging) as a first-class diagram. Pointing at a shared-responsibility matrix draws a deficiency.
Network and RF testing won't cover what matters: web app, API, mobile binary, OAuth/SSO redirection, multi-tenant data isolation, and cloud config drift. Scope must cite OWASP ASVS / MASVS.
Read: SaMD vs SiMD, what manufacturers need to knowBecause you control the production environment, reviewers expect continuous deployment with documented change control and a CVE-to-patch SLA in days-to-weeks, not the quarterly cycle a SiMD can justify.
If your SaMD includes a model (clinical decision support, image analysis, risk scoring, LLM features), reviewers expect a PCCP, GMLP-aligned lifecycle, adversarial robustness evidence, and transparency labeling on top of the standard Section 524B package. We bundle the AI/ML extension into the SaMD engagement so nothing falls between services.
Scoped to your modality: cloud-hosted CDS, mobile diagnostic app, web image-analysis viewer, digital therapeutic, or AI/ML SaMD.
Scope first, then test what matters, then document for the FDA and the notified body.
STRIDE-based threat model of the cloud architecture, mobile/web client, CI/CD, and IAM, with explicit RACI for manufacturer, cloud provider, and operating organization per IEC 82304-1.
Cybersecurity risk assessment (AAMI SW96), security architecture views, SBOM (app + container), patchability narrative, and operating-environment assumptions for labeling.
Manual-led web/API pen test (OWASP ASVS), mobile pen test (MASVS) where applicable, cloud configuration review, and IAM/tenancy isolation validation. Reproducible artifacts.
FDA-reviewer-ready evidence binder mapped to the 2026 premarket guidance and Section 524B, plus a parallel pack aligned to MDCG 2019-16 / IEC 81001-5-1 for EU MDR.
Postmarket monitoring (continuous SBOM monitoring, CSPM drift alerts, CVE-to-patch SLA, CVD intake) is a separate annual contract. You can bundle it with premarket at signature so the same team carries you from kickoff through your first year on market under one fixed fee.
Every samd cybersecurity engagement ships with the artifacts FDA reviewers expect to see - traceable, complete, and aligned with current guidance.
Every samd cybersecurity engagement produces evidence aligned to the regulatory and consensus standards FDA reviewers and notified bodies expect to see - traceable, complete, and ready to drop into your ISO 13485 quality system.
Defines the SPDF, Section 524B submission package, threat modeling, SBOM, security architecture views, and cybersecurity testing every cyber device submission must include.
Statutory requirement that every cyber device 510(k), De Novo, PMA, and IDE submission include a complete cybersecurity package or face Refuse to Accept (RTA).
End-to-end secure development lifecycle the FDA expects to see referenced and evidenced in every cyber device submission.
International standard for security activities across the health software product lifecycle.
The consensus standard for medical device security risk management - asset, threat, vulnerability, likelihood, severity, and residual risk acceptability.
Foundational risk management standard. Cybersecurity risk is tied directly to patient-safety risk in the 14971 file.
Verification requirements for web and application security controls.
Verification requirements for iOS / Android companion app security controls.
The most common and impactful API security risks - required reading for any device with cloud or partner APIs.
A senior medical device cybersecurity advisor on your team every week, for a fixed monthly fee.
View MedTech Cybersecurity AdvisoryOne package for applicable FDA cybersecurity documentation, threat modeling, SBOMs, testing, and submission support.
View FDA Premarket Cybersecurity PackageCybersecurity content for Predetermined Change Control Plans, for AI/ML and non-AI devices.
View PCCP Cybersecurity ServicesPressure-test the work yourself before you scope an engagement. No signup, results are yours to keep.
We own 100% of the FDA Section 524B premarket cybersecurity submission for your SaMD: threat model, secure architecture, SBOM, cloud + mobile + API pen testing, vulnerability management plan, labeling, and reviewer-ready evidence binder. Most modern SaMD includes AI/ML, so we extend the same package with PCCP, GMLP, adversarial robustness, and transparency labeling when your device uses ML. Postmarket monitoring is a separate annual contract, but you can bundle it with premarket at signature for a single fixed fee from kickoff through your first year on market.
Pick the shape that looks like your device. Same fixed-fee package, scoped to the attack surface that actually applies.
Multi-tenant SaaS that ingests clinical data, runs decision logic, and writes back to the EHR. We cover IAM, KMS, tenancy isolation, OWASP ASVS, and CSPM drift baselines.
Native iOS / Android SaMD where the phone is the device. We cover OWASP MASVS, certificate pinning, runtime tampering, code-signing, and the backend API the app talks to.
Browser-based viewer for radiology, pathology, or wound imaging. We cover OAuth/SSO redirection, DICOM transport, session management, and clinical-workflow business-logic abuse.
ML clinical decision support, image analysis, risk scoring, or LLM features. We layer PCCP, GMLP, adversarial robustness, and transparency labeling on top of the standard Section 524B package.
The same SaMD artifacts satisfy the FDA Section 524B, AAMI SW96, IEC 81001-5-1, and MDCG 2019-16 - mapped row-by-row so you don't pay to rebuild evidence for each filing. This is the table reviewers cite when they ask "where is your control for X?"
| Deliverable | FDA §524B + 2026 guidance | AAMI SW96 ANSI/AAMI | IEC 81001-5-1 lifecycle activities | MDCG 2019-16 EU MDR notified body |
|---|---|---|---|---|
Cybersecurity threat model (STRIDE) Cloud architecture, mobile/web client, CI/CD, IAM and tenancy boundaries. | Primary | Primary | Primary | Primary |
Cybersecurity risk assessment Patient-safety risk linkage to ISO 14971; AAMI SW96 risk-acceptability rationale. | Primary | Primary | Primary | Primary |
Secure design & architecture views Reference architecture, trust boundaries, data flows, IAM/KMS, multi-tenant isolation. | Primary | Supports | Primary | Primary |
SBOM (application + container) CycloneDX, third-party + transitive dependencies, container base images. | Primary | Supports | Primary | Primary |
VEX statements Exploitability per CVE in the deployed configuration; published alongside advisories. | Primary | Supports | Supports | Primary |
Penetration testing (web / API / mobile / cloud) OWASP ASVS L2/L3, OWASP API Top 10, OWASP MASVS, cloud configuration review. | Primary | Primary | Primary | Primary |
Vulnerability management plan CVE-to-patch SLA, change control, CVD intake, postmarket monitoring procedures. | Primary | Primary | Primary | Primary |
Coordinated vulnerability disclosure (CVD) Published security.txt, disclosure inbox, triage workflow, FDA postmarket reporting timelines. | Primary | Supports | Primary | Primary |
Operating-environment & accompanying documentation IEC 82304-1 operating-environment statement; cybersecurity labeling for users. | Primary | Supports | Primary | Primary |
AI/ML extension (PCCP, GMLP, adversarial robustness) Layered when the SaMD includes a model; references AAMI CR34971 and NIST AI RMF. | Primary | Supports | Supports | Supports |
Primary = framework explicitly requires this artifact. Supports = framework accepts it as evidence or references it indirectly.
Most SaMD founders compare us against either a boutique app-security shop or a Big-4 medical-device practice. Here's how the three options actually stack up on what reviewers care about.
| Dimension | Blue Goat Cyber SaMD specialists | Boutique pen-test shop app-security focus | Big consultancy Big-4 / large MedTech advisory |
|---|---|---|---|
| SaMD-specific scope (cloud + mobile + API + AI/ML) | Built for SaMD | Often app-only | Generic medical-device template |
| Senior engineer leads from minute one | Yes | Yes | Tiered delivery model |
| Fixed fee in 24 hours, no T&M billing | Yes | Often hourly | T&M default |
| Reviewer-ready FDA evidence binder included | Mapped to 2026 guidance | Pen test report only | Doc team is separate scope |
| AI/ML extension (PCCP, GMLP, adversarial robustness) bundled | Yes | No | Separate AI practice |
| Postmarket bundle option (annual SBOM/CVD/CSPM) | 1-yr bundle at signature | No | Different team |
| Typical cycle time, kickoff to submission | 6 to 10 weeks | Test only, ~3 wks | 12 to 20+ weeks |
| no-cost deficiency resolution commitment on cybersecurity deficiencies | Yes | No | No |
Generalized comparison based on common engagement structures we encounter in scoping calls. Individual firms may differ.
Cycle time and outcome from recent SaMD engagements. Device names withheld under NDA; modality, class, and the actual evidence binder are available under MNDA on request.
Full Section 524B package: STRIDE threat model of the cloud architecture, OWASP ASVS web/API pen test, CycloneDX SBOM, IAM and tenancy isolation review, reviewer-ready evidence binder.
8 weeks · kickoff to submission
Cleared with zero cybersecurity deficiencies
OWASP MASVS mobile pen test on both platforms, backend API testing, code-signing and pinning review, IEC 82304-1 operating-environment labeling, MDCG 2019-16 parallel pack for EU MDR.
7 weeks · kickoff to submission
Cleared with one minor information request, resolved in 5 days
Section 524B core package plus AI/ML extension: PCCP, GMLP-aligned lifecycle, adversarial robustness evidence (FGSM + PGD), AAMI CR34971 risk extension, transparency labeling for the ML model card.
10 weeks · kickoff to submission
Cleared with PCCP accepted as filed
We don't bill hourly and we don't sell tiers. Every SaMD engagement is a single fixed fee, quoted in writing within 24 hours of a free 30-minute discovery call. The fee tracks the actual attack surface of your device and the submission window you're working against - not seat counts, not retainers, not add-ons.
SaMD engagements are typically lower than comparable connected-hardware projects because there is no firmware or RF testing scope. Postmarket is a separate annual contract; bundle pricing is quoted alongside.
Backed by our no-cost deficiency resolution commitment: if a cybersecurity deficiency in our deliverables delays your submission, we resolve it at no additional cost until clearance.
Background reading and adjacent services for teams scoping a SaMD premarket submission.
The full reference: what the FDA and notified bodies expect for Software as a Medical Device, mapped to Section 524B, IEC 81001-5-1, and MDCG 2019-16.
Read the guideWhy the distinction drives your FDA strategy, which standards apply to each, and how cybersecurity scope changes when there's no firmware to test.
Read the postWhen your SaMD includes a model, the package extends with PCCP, GMLP, adversarial robustness, and transparency labeling, layered on top of the Section 524B core.
See the AI/ML service
