<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: SecureSlate</title>
    <description>The latest articles on DEV Community by SecureSlate (@secureslate).</description>
    <link>https://dev.to/secureslate</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4247640%2Fe700d746-37a9-4e31-a150-2e5dad54355e.png</url>
      <title>DEV Community: SecureSlate</title>
      <link>https://dev.to/secureslate</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/secureslate"/>
    <language>en</language>
    <item>
      <title>Threat Modeling for HealthTech: A Practical Guide for Apps That Handle Patient Data</title>
      <dc:creator>SecureSlate</dc:creator>
      <pubDate>Mon, 16 Nov 2026 14:14:02 +0000</pubDate>
      <link>https://dev.to/secureslate/threat-modeling-for-healthtech-a-practical-guide-for-apps-that-handle-patient-data-c9g</link>
      <guid>https://dev.to/secureslate/threat-modeling-for-healthtech-a-practical-guide-for-apps-that-handle-patient-data-c9g</guid>
      <description>&lt;p&gt;&lt;strong&gt;Short answer:&lt;/strong&gt; Threat modeling for HealthTech is a structured design review that asks how patient data could be exposed, altered or made unavailable, and what you will do about it. It gives your HIPAA risk analysis concrete, system-level inputs, and for connected medical devices FDA's premarket cybersecurity guidance recommends submitting threat modeling documentation.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Related guides:&lt;/strong&gt; &lt;a href="https://getsecureslate.com/blog/learn-risk-stride-threat-model" rel="noopener noreferrer"&gt;STRIDE threat model&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/hipaa-risk-assessment" rel="noopener noreferrer"&gt;HIPAA risk assessment&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/fda-sbom-requirements-medical-devices" rel="noopener noreferrer"&gt;FDA SBOM requirements for medical devices&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Key takeaways
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Threat modeling is a &lt;strong&gt;design-time&lt;/strong&gt; activity. It finds problems while they are still cheap to fix, before code and contracts lock them in.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Every threat model answers &lt;strong&gt;four questions&lt;/strong&gt;: what are we working on, what can go wrong, what are we going to do about it, and did we do a good enough job.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;In HealthTech, the output feeds your &lt;strong&gt;HIPAA risk analysis&lt;/strong&gt;, your secure development evidence for &lt;strong&gt;SOC 2 and ISO 27001&lt;/strong&gt;, and, for cyber devices, your &lt;strong&gt;FDA premarket submission&lt;/strong&gt;.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Start small. One diagram, one hour and one product feature beats a perfect model that never gets finished.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What is threat modeling, in plain terms?
&lt;/h2&gt;

&lt;p&gt;Threat modeling is looking at a picture of your system and asking what could go wrong with its security and privacy. The &lt;a href="https://threatmodelingmanifesto.org/" rel="noopener noreferrer"&gt;Threat Modeling Manifesto&lt;/a&gt; defines it as "analyzing representations of a system to highlight concerns about security and privacy characteristics", and frames the work around four questions.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;What are we working on?&lt;/strong&gt; Usually a data flow diagram that shows users, services, data stores, third parties and the trust boundaries between them.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;What can go wrong?&lt;/strong&gt; The threats, found by walking the diagram with a method or checklist.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;What are we going to do about it?&lt;/strong&gt; Mitigate, accept, transfer or avoid each threat, with an owner.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Did we do a good enough job?&lt;/strong&gt; Check the model against reality and review it when the system changes.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;That is the whole idea. The methods described below are different ways of answering question two.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why does threat modeling matter more in HealthTech?
&lt;/h2&gt;

&lt;p&gt;It matters more because the data is sensitive, the regulations expect you to analyze risk, and some products are medical devices with their own premarket expectations. A breach of protected health information (PHI) hurts patients, triggers notification duties and damages trust with the health systems you sell to.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Requirement&lt;/th&gt;
&lt;th&gt;What it asks for&lt;/th&gt;
&lt;th&gt;How a threat model helps&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;HIPAA risk analysis, 45 CFR 164.308(a)(1)(ii)(A)&lt;/td&gt;
&lt;td&gt;An accurate and thorough assessment of potential risks and vulnerabilities to the confidentiality, integrity and availability of ePHI&lt;/td&gt;
&lt;td&gt;Supplies system-specific threats and data flows instead of generic risk statements&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SOC 2 CC3.2, Trust Services Criteria&lt;/td&gt;
&lt;td&gt;Identify risks to objectives and analyze them as a basis for managing them&lt;/td&gt;
&lt;td&gt;Shows how product risks were identified and treated&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ISO 27001:2022, Annex A 8.25 and 8.27&lt;/td&gt;
&lt;td&gt;Secure development life cycle and secure system architecture and engineering principles&lt;/td&gt;
&lt;td&gt;Provides evidence that security is considered during design&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;FDA premarket cybersecurity guidance&lt;/td&gt;
&lt;td&gt;For devices in scope, FDA recommends premarket submissions include threat modeling documentation&lt;/td&gt;
&lt;td&gt;The model and its mitigations become part of the submission&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;On the medical device side, FDA's guidance &lt;a href="https://www.fda.gov/media/119933/download" rel="noopener noreferrer"&gt;Cybersecurity in Medical Devices: Quality Management System Considerations and Content of Premarket Submissions&lt;/a&gt;, in the version issued in February 2026, recommends that threat modeling be performed throughout the design process and included in premarket submissions. Section 524B of the Federal Food, Drug, and Cosmetic Act sets cybersecurity requirements for "cyber devices", which the guidance describes as devices that include sponsor-authorized software, can connect to the internet, and have characteristics that could be vulnerable to cybersecurity threats. If your product may be a medical device, confirm its regulatory status with qualified advisors early. Our guide to &lt;a href="https://getsecureslate.com/blog/fda-sbom-requirements-medical-devices" rel="noopener noreferrer"&gt;FDA SBOM requirements&lt;/a&gt; covers another part of the same submission.&lt;/p&gt;

&lt;p&gt;See SecureSlate map one control across SOC 2, ISO 27001 and HIPAA&lt;/p&gt;

&lt;h2&gt;
  
  
  Which threat modeling method should you use?
&lt;/h2&gt;

&lt;p&gt;Use the method your team will actually run. For most HealthTech software teams that means STRIDE for security threats, plus a privacy lens for PHI. The table compares common options.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Method&lt;/th&gt;
&lt;th&gt;Focus&lt;/th&gt;
&lt;th&gt;Good fit when&lt;/th&gt;
&lt;th&gt;Effort&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;STRIDE&lt;/td&gt;
&lt;td&gt;Six security threat categories applied to each element of a diagram&lt;/td&gt;
&lt;td&gt;You want a quick, repeatable checklist for engineers&lt;/td&gt;
&lt;td&gt;Low to moderate&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;PASTA&lt;/td&gt;
&lt;td&gt;Seven-stage, risk-centric process tied to business impact&lt;/td&gt;
&lt;td&gt;You need to connect threats to business risk and attack simulation&lt;/td&gt;
&lt;td&gt;High&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;LINDDUN&lt;/td&gt;
&lt;td&gt;Privacy threats such as linkability and identifiability&lt;/td&gt;
&lt;td&gt;You process PHI or other personal data and want a privacy view&lt;/td&gt;
&lt;td&gt;Moderate&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Attack trees&lt;/td&gt;
&lt;td&gt;A goal at the root, ways to reach it as branches&lt;/td&gt;
&lt;td&gt;You want to explore one high-impact scenario in depth&lt;/td&gt;
&lt;td&gt;Variable&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;See our explainers on &lt;a href="https://getsecureslate.com/blog/learn-risk-stride-threat-model" rel="noopener noreferrer"&gt;the STRIDE threat model&lt;/a&gt; and &lt;a href="https://getsecureslate.com/blog/learn-risk-pasta-threat-modeling" rel="noopener noreferrer"&gt;PASTA threat modeling&lt;/a&gt; for how each one works. For connected medical devices, the &lt;a href="https://www.mitre.org/sites/default/files/2021-11/Playbook-for-Threat-Modeling-Medical-Devices.pdf" rel="noopener noreferrer"&gt;Playbook for Threat Modeling Medical Devices&lt;/a&gt;, developed by a team including MITRE and the Medical Device Innovation Consortium (MDIC), is a useful free reference.&lt;/p&gt;

&lt;h2&gt;
  
  
  What does a threat model look like for a telehealth app?
&lt;/h2&gt;

&lt;p&gt;A worked example makes the idea concrete. Imagine a telehealth product with a patient mobile app, a clinician web portal, an API, a database of visit notes, a third-party video provider, an EHR integration and a product analytics tool.&lt;/p&gt;

&lt;p&gt;Walking that diagram produces threats like these. They are illustrative, not a complete model.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Example threat&lt;/th&gt;
&lt;th&gt;Where it sits&lt;/th&gt;
&lt;th&gt;Example mitigation&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;A patient changes a record ID in an API call and sees another patient's notes&lt;/td&gt;
&lt;td&gt;API to database&lt;/td&gt;
&lt;td&gt;Object-level authorization checks on every request, with automated tests&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;PHI ends up in analytics events through screen names or form fields&lt;/td&gt;
&lt;td&gt;Mobile app to analytics tool&lt;/td&gt;
&lt;td&gt;Strip identifiers before events are sent, and review the vendor's terms and BAA position&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;A clinician session token is stolen from a shared workstation&lt;/td&gt;
&lt;td&gt;Clinician portal&lt;/td&gt;
&lt;td&gt;Short session lifetimes, re-authentication for sensitive actions, phishing-resistant MFA&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;The video provider processes PHI without a business associate agreement&lt;/td&gt;
&lt;td&gt;Trust boundary to vendor&lt;/td&gt;
&lt;td&gt;Confirm a BAA is signed before go-live and record it in your vendor inventory&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Application logs capture visit notes in plain text&lt;/td&gt;
&lt;td&gt;API to logging service&lt;/td&gt;
&lt;td&gt;Log redaction rules and restricted log access&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;The EHR integration credentials are hard-coded in the repository&lt;/td&gt;
&lt;td&gt;Integration service&lt;/td&gt;
&lt;td&gt;Secrets manager, rotation, and secrets scanning in CI&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Notice how many of these threats sit at &lt;strong&gt;trust boundaries&lt;/strong&gt;: between the patient and your API, between your systems and a vendor, between production data and logs. Drawing the boundaries is often where teams find their biggest risks.&lt;/p&gt;

&lt;p&gt;How audit-ready are you? Answer 12 questions and get your readiness score&lt;/p&gt;

&lt;h2&gt;
  
  
  How do you run your first threat modeling session?
&lt;/h2&gt;

&lt;p&gt;Pick one feature, invite the people who built it, and timebox the session to about an hour. A simple agenda:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Scope.&lt;/strong&gt; Choose a single feature or service, ideally a new one that touches PHI.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Diagram.&lt;/strong&gt; Draw the data flow on a whiteboard: actors, processes, data stores, external services and trust boundaries.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Mark the PHI.&lt;/strong&gt; Highlight every flow and store that carries ePHI.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Find threats.&lt;/strong&gt; Walk each element and boundary with your chosen method. Write threats as short sentences: "an attacker could..."&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Rate and decide.&lt;/strong&gt; Give each threat a rough likelihood and impact, then choose a response and an owner.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Track.&lt;/strong&gt; Turn mitigations into tickets and add the residual risks to your risk register.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Store the model.&lt;/strong&gt; Save the diagram, threat list and decisions with a date so you can show them later.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Keep the first one small. In our experience, teams that try to model the whole platform in one go often stall. Bake the habit into your &lt;a href="https://getsecureslate.com/blog/secure-sdlc-policy" rel="noopener noreferrer"&gt;secure SDLC policy&lt;/a&gt; so new features get modeled as part of design review.&lt;/p&gt;

&lt;h2&gt;
  
  
  When should you update a threat model?
&lt;/h2&gt;

&lt;p&gt;Update a threat model whenever the system changes in a way that affects data flows or trust boundaries, and review it on a regular schedule. Typical triggers include:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;A new feature that collects or displays PHI&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;A new third-party service, SDK or integration&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;A change to authentication, hosting or network architecture&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;A security incident or a significant penetration test finding&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Your annual HIPAA risk analysis, so the two stay consistent&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;A model that matches last year's architecture gives false comfort. Small, frequent updates are easier than rare rewrites.&lt;/p&gt;

&lt;h2&gt;
  
  
  How SecureSlate helps
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;SecureSlate&lt;/a&gt; helps HealthTech teams turn threat modeling outputs into tracked risks and audit-ready evidence:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Risk management&lt;/strong&gt; to record threats, owners, ratings and treatment decisions in one register.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Built-in HIPAA, SOC 2 and ISO 27001 frameworks&lt;/strong&gt; with multi-framework control mapping, so design-review evidence supports all three.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Code security and secrets detection&lt;/strong&gt; to catch issues such as hard-coded credentials that threat models often flag.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Vendor risk management&lt;/strong&gt; to track third parties that sit across your trust boundaries, including BAA status.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Audit management&lt;/strong&gt; to keep dated threat models and decisions ready for auditors and customer reviews.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;SecureSlate does not prepare FDA submissions or determine whether your product is a medical device.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;Start a SecureSlate trial&lt;/a&gt; to connect your threat models to your risk register.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Does HIPAA require threat modeling?
&lt;/h3&gt;

&lt;p&gt;HIPAA does not name threat modeling. It requires an accurate and thorough risk analysis of potential risks and vulnerabilities to ePHI. A threat model is a practical way to produce system-level inputs for that analysis, especially for software products.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is STRIDE enough for a HealthTech product?
&lt;/h3&gt;

&lt;p&gt;STRIDE covers core security threats well, but it was not designed specifically for privacy. Many HealthTech teams pair it with a privacy-focused review of how PHI is collected, linked and shared, for example using LINDDUN or a simple PHI data-flow check.&lt;/p&gt;

&lt;h3&gt;
  
  
  Who should attend a threat modeling session?
&lt;/h3&gt;

&lt;p&gt;Include the engineers who build the feature, someone who understands the architecture, and a security or compliance lead. For features that handle PHI, a product owner who knows how clinicians and patients use the feature is also valuable.&lt;/p&gt;

&lt;h3&gt;
  
  
  How long does threat modeling take?
&lt;/h3&gt;

&lt;p&gt;A focused session on one feature can take about an hour once the team is familiar with the process. Larger systems or medical device submissions take longer because they need fuller documentation and traceability to mitigations.&lt;/p&gt;

&lt;h2&gt;
  
  
  Disclaimer (legal note)
&lt;/h2&gt;

&lt;p&gt;This article is for general information only and is not legal, regulatory, or professional advice. Requirements vary by framework, industry, and jurisdiction. Consult qualified advisors for your specific obligations.&lt;/p&gt;

&lt;h3&gt;
  
  
  Need compliance without the complexity?
&lt;/h3&gt;

&lt;p&gt;SecureSlate automates ISO 27001, SOC 2, GDPR, HIPAA, and more. Built for growing teams. See it in action.&lt;/p&gt;

&lt;p&gt;Learn more: &lt;a href="https://getsecureslate.com/frameworks/hipaa" rel="noopener noreferrer"&gt;HIPAA compliance with SecureSlate&lt;/a&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://getsecureslate.com/blog/threat-modeling-for-healthtech" rel="noopener noreferrer"&gt;SecureSlate blog&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>security</category>
      <category>healthtech</category>
      <category>architecture</category>
      <category>webdev</category>
    </item>
    <item>
      <title>Employee Offboarding Security Checklist: Revoke Access Before It Becomes an Audit Finding</title>
      <dc:creator>SecureSlate</dc:creator>
      <pubDate>Mon, 16 Nov 2026 14:07:38 +0000</pubDate>
      <link>https://dev.to/secureslate/employee-offboarding-security-checklist-revoke-access-before-it-becomes-an-audit-finding-3dce</link>
      <guid>https://dev.to/secureslate/employee-offboarding-security-checklist-revoke-access-before-it-becomes-an-audit-finding-3dce</guid>
      <description>&lt;p&gt;&lt;strong&gt;Short answer:&lt;/strong&gt; An employee offboarding security checklist makes sure every account, credential, device and copy of sensitive data is removed or recovered when someone leaves. Disable the identity provider account first, then revoke tokens, keys and non-SSO apps, collect devices, and transfer data ownership. Record each step with a timestamp so you can prove it to an auditor.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Related guides:&lt;/strong&gt; &lt;a href="https://getsecureslate.com/blog/personnel-management-compliance-guide" rel="noopener noreferrer"&gt;Personnel management for compliance&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/device-offboarding-with-mdm-prevent-orphaned-endpoints" rel="noopener noreferrer"&gt;device offboarding with MDM&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/how-to-perform-quarterly-access-reviews" rel="noopener noreferrer"&gt;how to perform quarterly access reviews&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Key takeaways
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;HIPAA, SOC 2 and ISO 27001 all expect you to &lt;strong&gt;remove access when someone leaves&lt;/strong&gt;, and to be able to show you did it.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;The identity provider is the off switch for most systems, but &lt;strong&gt;not all of them&lt;/strong&gt;. API keys, personal access tokens, local accounts and shared passwords survive an SSO shutdown.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Treat &lt;strong&gt;role changes and contractors&lt;/strong&gt; with the same rigor as departures. Access that outlives the job that needed it is the same risk.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Your evidence is the &lt;strong&gt;ticket trail&lt;/strong&gt;: who asked, what was removed, when, and a later reconciliation that proves nothing was missed.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why is offboarding a security risk?
&lt;/h2&gt;

&lt;p&gt;Offboarding is a security risk because every account, token or device that is not removed becomes access nobody is watching. A former employee usually means no harm, but a dormant account with a valid password is still an attack path, and in a HealthTech company it may lead straight to electronic protected health information (ePHI).&lt;/p&gt;

&lt;p&gt;In our experience, the gaps are rarely in the main SSO account. They hide in the places teams forget to list:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;SaaS tools signed up for with a password instead of SSO&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Personal access tokens in source control and CI systems&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Cloud IAM users and long-lived access keys&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Shared admin passwords and break-glass accounts&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Data synced to a personal laptop or phone&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Calendar delegation, mail forwarding rules and shared drive ownership&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;A checklist that names these explicitly is the difference between "we disabled the account" and "we removed the access".&lt;/p&gt;

&lt;h2&gt;
  
  
  What do HIPAA, SOC 2 and ISO 27001 expect at offboarding?
&lt;/h2&gt;

&lt;p&gt;All three expect access to be removed when it is no longer needed, but they describe it in different words. The table maps the main requirements so one process can satisfy each of them.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Framework&lt;/th&gt;
&lt;th&gt;Relevant requirement&lt;/th&gt;
&lt;th&gt;What it asks for&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;HIPAA Security Rule&lt;/td&gt;
&lt;td&gt;45 CFR 164.308(a)(3)(ii)(C), termination procedures (addressable)&lt;/td&gt;
&lt;td&gt;Procedures for terminating access to ePHI when a workforce member's employment or other arrangement ends&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;HIPAA Security Rule&lt;/td&gt;
&lt;td&gt;45 CFR 164.310(d)(1), device and media controls&lt;/td&gt;
&lt;td&gt;Policies for the receipt and removal of hardware and electronic media that contain ePHI&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SOC 2&lt;/td&gt;
&lt;td&gt;Trust Services Criteria CC6.2 and CC6.3&lt;/td&gt;
&lt;td&gt;Credentials removed when access is no longer authorized, and access modified or removed based on roles and responsibilities&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ISO 27001:2022&lt;/td&gt;
&lt;td&gt;Annex A controls 5.11, 5.18 and 6.5&lt;/td&gt;
&lt;td&gt;Return of assets, management of access rights, and responsibilities that continue after termination or change of employment&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;"Addressable" under HIPAA does not mean optional. It means you assess whether the specification is reasonable and appropriate for your environment, and document an equivalent measure if you choose a different approach. For almost every organization that handles ePHI, a documented termination procedure is the reasonable choice.&lt;/p&gt;

&lt;p&gt;A note on regulatory change: in its &lt;a href="https://www.hhs.gov/hipaa/for-professionals/security/hipaa-security-rule-nprm/factsheet/index.html" rel="noopener noreferrer"&gt;proposed update to the HIPAA Security Rule&lt;/a&gt;, HHS proposed requiring notification of certain regulated entities within 24 hours when a workforce member's access to ePHI or certain electronic information systems is changed or terminated. As of November 2026 this is a proposal, not a final rule, but it shows the direction of travel. A fast, documented offboarding process prepares you either way.&lt;/p&gt;

&lt;p&gt;For a broader control reference, NIST's &lt;a href="https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final" rel="noopener noreferrer"&gt;SP 800-53&lt;/a&gt; control PS-4 (personnel termination) asks organizations to disable system access within a time period they define, revoke credentials and retrieve organizational property.&lt;/p&gt;

&lt;p&gt;See SecureSlate map one control across SOC 2, ISO 27001 and HIPAA&lt;/p&gt;

&lt;h2&gt;
  
  
  What should happen before, on and after the last day?
&lt;/h2&gt;

&lt;p&gt;Split offboarding into four windows so each task has an owner and a deadline. The timings below are examples. Set your own in policy and keep to them.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Window&lt;/th&gt;
&lt;th&gt;Goal&lt;/th&gt;
&lt;th&gt;Typical owner&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Before the last day&lt;/td&gt;
&lt;td&gt;Know everything the person can reach&lt;/td&gt;
&lt;td&gt;IT or security, with the manager&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Last day (or immediately, for involuntary exits)&lt;/td&gt;
&lt;td&gt;Cut off access and recover devices&lt;/td&gt;
&lt;td&gt;IT, with HR&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;First week&lt;/td&gt;
&lt;td&gt;Close the long tail: tokens, shared secrets, data ownership&lt;/td&gt;
&lt;td&gt;IT, engineering, system owners&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Within 30 days&lt;/td&gt;
&lt;td&gt;Prove nothing was missed&lt;/td&gt;
&lt;td&gt;Security or compliance&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Before the last day.&lt;/strong&gt; Pull an access inventory for the person from your identity provider, cloud accounts, code platforms and any app that does not use SSO. Ask the manager which shared accounts or secrets the person knew. Plan the handover of files, recurring jobs and customer relationships.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;On the last day.&lt;/strong&gt; Disable the identity provider account and end active sessions. Remove the person from admin groups. Collect laptops, phones, hardware keys and badges. If they used a personal device for work, follow your policy for removing company data from it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;In the first week.&lt;/strong&gt; Revoke tokens and keys, rotate shared secrets they knew, transfer document ownership, and remove mail forwarding rules. Check that scheduled jobs or integrations do not run under their identity.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Within 30 days.&lt;/strong&gt; Reconcile the HR leaver list against active accounts across your systems. Anything still active is a finding to fix and explain.&lt;/p&gt;

&lt;h2&gt;
  
  
  The employee offboarding security checklist
&lt;/h2&gt;

&lt;p&gt;Use this as a starting template and tailor it to your stack.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Identity and access&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Disable the identity provider account and revoke active sessions&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Remove or reset multi-factor authentication factors and hardware keys&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Remove from privileged and admin groups&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Disable accounts in apps that do not use SSO&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Remove from shared vaults and change any shared passwords they knew&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Cloud and infrastructure&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Delete or disable cloud IAM users and access keys&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Remove SSH keys from servers and bastion hosts&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Remove VPN and database accounts&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Check that no service accounts or automations depend on their credentials&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Code and pipelines&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Remove from code hosting organizations and teams&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Revoke personal access tokens and deploy keys they created&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Rotate secrets stored in CI variables they could read&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Scan for hard-coded credentials tied to the person&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Devices and data&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Recover company laptops, phones and removable media&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Lock or wipe managed devices through your device management tool&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Remove company data and ePHI from any personal device used for work&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Transfer ownership of files, shared drives and recurring meetings&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;People and paperwork&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Remind the person of confidentiality obligations that continue after they leave&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Record the exit date, request time and completion time for each step&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Notify customers or partners whose systems the person could access, where contracts require it&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If the person had access to ePHI, add a check that their removal is reflected in your &lt;a href="https://getsecureslate.com/blog/hipaa-risk-assessment" rel="noopener noreferrer"&gt;HIPAA risk analysis&lt;/a&gt; records of who can access which systems. The same logic applies to departing vendors, covered in our guide to &lt;a href="https://getsecureslate.com/blog/vendor-offboarding-best-practices-for-reducing-risk" rel="noopener noreferrer"&gt;vendor offboarding&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;How audit-ready are you? Answer 12 questions and get your readiness score&lt;/p&gt;

&lt;h2&gt;
  
  
  Do role changes and contractors need the same process?
&lt;/h2&gt;

&lt;p&gt;Yes. A person who changes role, or a contractor whose engagement ends, needs the same discipline even though nobody hands back a laptop. SOC 2 CC6.3 covers modifying and removing access based on roles, and ISO 27001 control 6.5 explicitly covers changes of employment, not only terminations.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Leavers:&lt;/strong&gt; run the full checklist.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Movers:&lt;/strong&gt; add the access the new role needs, then remove what the old role had. The common failure is adding without removing, which slowly builds up excessive access.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Contractors and agency staff:&lt;/strong&gt; set an end date on their accounts when you create them, and tie access to a named internal sponsor who confirms when the work ends.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What evidence will an auditor ask for?
&lt;/h2&gt;

&lt;p&gt;Auditors want to see a defined process, proof it ran for each leaver in the audit period, and a check that catches misses. Expect requests like these:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Your offboarding policy or procedure,&lt;/strong&gt; including the timeframe you commit to for removing access. SOC 2 does not set a number, so pick one you can meet consistently.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;A population of leavers&lt;/strong&gt; from HR for the audit period.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Samples from that population,&lt;/strong&gt; each with a ticket or record showing the request date, the systems removed and the completion time.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Account reconciliations&lt;/strong&gt; showing that leavers no longer have active accounts in key systems, often as part of &lt;a href="https://getsecureslate.com/blog/how-to-perform-quarterly-access-reviews" rel="noopener noreferrer"&gt;quarterly access reviews&lt;/a&gt;.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Device return records&lt;/strong&gt; for company hardware.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;In our experience, the most common exception is timing: access removed, but days after the termination date with no explanation. A clear policy deadline and an automated trigger from HR to IT closes most of that gap.&lt;/p&gt;

&lt;h2&gt;
  
  
  How SecureSlate helps
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;SecureSlate&lt;/a&gt; helps HealthTech and SaaS teams run offboarding as a control, not a scramble:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Agentless read-only cloud integrations&lt;/strong&gt; to see which accounts exist across your identity provider, cloud and SaaS tools.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Access reviews&lt;/strong&gt; to reconcile leavers and movers against active accounts and record the result.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Device agent&lt;/strong&gt; to check the security posture of company devices before and after they are returned.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Code security and secrets detection&lt;/strong&gt; to find credentials in repositories that should be rotated.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Multi-framework control mapping&lt;/strong&gt; across built-in HIPAA, SOC 2 and ISO 27001, so one offboarding control serves all three.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Audit management&lt;/strong&gt; to keep tickets, reconciliations and timestamps ready for your auditor.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;Start a SecureSlate trial&lt;/a&gt; to map your offboarding control across frameworks.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  How quickly should access be removed when an employee leaves?
&lt;/h3&gt;

&lt;p&gt;HIPAA, SOC 2 and ISO 27001 do not set a fixed number of hours. Define a timeframe in your policy that fits your risk, and remove access immediately for involuntary terminations. Many teams aim for the same business day for core systems, then close the remaining items within a short, documented window.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is disabling the SSO account enough?
&lt;/h3&gt;

&lt;p&gt;Usually not. SSO shuts off apps that depend on it, but API keys, personal access tokens, local accounts, shared passwords and data on personal devices can remain. Your checklist should cover each of these explicitly.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is the HIPAA termination procedure required if it is addressable?
&lt;/h3&gt;

&lt;p&gt;Addressable means you must assess whether it is reasonable and appropriate and document your decision, including any alternative measure. It does not mean you can skip it without analysis. For organizations handling ePHI, a documented termination procedure is almost always appropriate.&lt;/p&gt;

&lt;h3&gt;
  
  
  Does offboarding apply to contractors under SOC 2?
&lt;/h3&gt;

&lt;p&gt;Yes. SOC 2 CC6.2 and CC6.3 apply to users whose access the organization administers, which typically includes contractors. Give contractor accounts an end date and a sponsor who confirms when the engagement finishes.&lt;/p&gt;

&lt;h2&gt;
  
  
  Disclaimer (legal note)
&lt;/h2&gt;

&lt;p&gt;This article is for general information only and is not legal, regulatory, or professional advice. Requirements vary by framework, industry, and jurisdiction. Consult qualified advisors for your specific obligations.&lt;/p&gt;

&lt;h3&gt;
  
  
  Need compliance without the complexity?
&lt;/h3&gt;

&lt;p&gt;SecureSlate automates ISO 27001, SOC 2, GDPR, HIPAA, and more. Built for growing teams. See it in action.&lt;/p&gt;

&lt;p&gt;Learn more: &lt;a href="https://getsecureslate.com/products/risk-management" rel="noopener noreferrer"&gt;Risk management with SecureSlate&lt;/a&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://getsecureslate.com/blog/employee-offboarding-security-checklist" rel="noopener noreferrer"&gt;SecureSlate blog&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>security</category>
      <category>devops</category>
      <category>productivity</category>
      <category>cloud</category>
    </item>
    <item>
      <title>EU Cyber Resilience Act: What Software Teams Must Do and By When</title>
      <dc:creator>SecureSlate</dc:creator>
      <pubDate>Mon, 16 Nov 2026 14:06:42 +0000</pubDate>
      <link>https://dev.to/secureslate/eu-cyber-resilience-act-what-software-teams-must-do-and-by-when-1n1l</link>
      <guid>https://dev.to/secureslate/eu-cyber-resilience-act-what-software-teams-must-do-and-by-when-1n1l</guid>
      <description>&lt;p&gt;&lt;strong&gt;Short answer:&lt;/strong&gt; The EU Cyber Resilience Act (Regulation (EU) 2024/2847) sets mandatory cybersecurity rules for hardware and software sold in the EU. Manufacturers must build products securely, handle vulnerabilities through a support period and carry CE marking. Reporting duties applied from 11 September 2026, and most other obligations apply from 11 December 2027.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Related guides:&lt;/strong&gt; &lt;a href="https://getsecureslate.com/blog/eu-cyber-resilience-act-sbom" rel="noopener noreferrer"&gt;EU Cyber Resilience Act SBOM requirements&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/eu-cybersecurity-regulations-for-saas" rel="noopener noreferrer"&gt;EU cybersecurity regulations for SaaS&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/vulnerability-management" rel="noopener noreferrer"&gt;vulnerability management&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Key takeaways
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;The CRA is a product law. It attaches to what you place on the EU market, not to your company's sector or size.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;As of November 2026, reporting of actively exploited vulnerabilities and severe incidents is already live, including for products sold before the full rules begin.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Pure SaaS is generally handled under NIS 2, but remote data processing that a product needs in order to work falls under the CRA.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Most products can use self-assessment. Important and critical products face tighter routes.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Treat 11 December 2027 as the date your secure development, documentation and support commitments must be provable.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What is the EU Cyber Resilience Act?
&lt;/h2&gt;

&lt;p&gt;The Cyber Resilience Act is an EU regulation that makes cybersecurity a condition for selling products with digital elements in the single market. It entered into force on 10 December 2024 and works much like other EU product safety law: meet the essential requirements, prove it, and mark the product.&lt;/p&gt;

&lt;p&gt;Before the CRA, security expectations for connected products depended on the product type or the customer contract. The regulation sets one baseline across the EU, covering how a product is designed and how its vulnerabilities are handled after release. The &lt;a href="https://digital-strategy.ec.europa.eu/en/policies/cra-summary" rel="noopener noreferrer"&gt;Commission's summary of the legal text&lt;/a&gt; is a useful plain-language companion to the &lt;a href="https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ%3AL_202402847" rel="noopener noreferrer"&gt;full regulation on EUR-Lex&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Responsibility is spread across the supply chain:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Role&lt;/th&gt;
&lt;th&gt;Core duty&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Manufacturer&lt;/td&gt;
&lt;td&gt;Meets the essential requirements, assesses conformity, handles vulnerabilities, reports incidents&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Authorised representative&lt;/td&gt;
&lt;td&gt;Performs delegated manufacturer tasks under a written mandate&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Importer&lt;/td&gt;
&lt;td&gt;Checks that non-EU products comply before placing them on the market&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Distributor&lt;/td&gt;
&lt;td&gt;Verifies CE marking and that earlier parties met their duties&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Open-source software steward&lt;/td&gt;
&lt;td&gt;Keeps a cybersecurity policy and cooperates with authorities, with lighter duties than a manufacturer&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Does the Cyber Resilience Act apply to your product?
&lt;/h2&gt;

&lt;p&gt;The CRA applies if you make a hardware or software product available on the EU market in a commercial activity and its intended or reasonably foreseeable use includes a data connection to a device or network. That wording is broad, and it reaches desktop and mobile software, firmware, connected devices and components sold separately.&lt;/p&gt;

&lt;p&gt;Work through these questions in order:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Is it supplied commercially in the EU?&lt;/strong&gt; Products not made available on the market fall outside the rules.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Does it have a data connection?&lt;/strong&gt; Direct or indirect, logical or physical, counts.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Is it covered by a sector law the CRA steps aside for?&lt;/strong&gt; &lt;a href="https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ%3AL_202402847" rel="noopener noreferrer"&gt;Article 2 of the regulation&lt;/a&gt; excludes medical devices under Regulations (EU) 2017/745 and 2017/746, vehicles under Regulation (EU) 2019/2144, certified aviation products, marine equipment, and products developed only for national security or defence.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Is it a cloud service?&lt;/strong&gt; &lt;a href="https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ%3AL_202402847" rel="noopener noreferrer"&gt;Recital 12&lt;/a&gt; notes that NIS 2 applies to cloud service models such as SaaS. Cloud solutions only count under the CRA when they meet its definition of remote data processing.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Remote data processing is the point many software companies miss. If your installed app or device cannot perform one of its functions without a backend that you design or control, that backend is part of the product for CRA purposes. A browser-only SaaS with no distributed product will usually look to &lt;a href="https://getsecureslate.com/blog/nis-2-compliance-checklist" rel="noopener noreferrer"&gt;NIS 2&lt;/a&gt; instead, but it is worth confirming this with counsel if you also ship agents, SDKs or desktop clients.&lt;/p&gt;

&lt;p&gt;HealthTech teams should check each product separately. Software regulated as a medical device under the EU medical device rules is outside the CRA, while a companion app or wellness product that is not a regulated device may not be.&lt;/p&gt;

&lt;p&gt;See SecureSlate map one control across SOC 2, ISO 27001 and HIPAA&lt;/p&gt;

&lt;h2&gt;
  
  
  When do the CRA obligations start?
&lt;/h2&gt;

&lt;p&gt;The CRA phases in over three years, and the first operational deadline has already passed. The &lt;a href="https://digital-strategy.ec.europa.eu/en/factpages/cyber-resilience-act-implementation" rel="noopener noreferrer"&gt;Commission's implementation page&lt;/a&gt; lists the milestones, and it showed no proposed changes to these dates when we checked in November 2026.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Date&lt;/th&gt;
&lt;th&gt;What applies&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;10 December 2024&lt;/td&gt;
&lt;td&gt;Regulation enters into force&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;11 June 2026&lt;/td&gt;
&lt;td&gt;Rules on notifying conformity assessment bodies apply&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;11 September 2026&lt;/td&gt;
&lt;td&gt;Article 14 reporting obligations apply&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;11 December 2027&lt;/td&gt;
&lt;td&gt;Main obligations apply in full&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Products placed on the market before 11 December 2027 only come under the wider rules if they are substantially modified after that date. Reporting is the exception, because it also covers products already on the market.&lt;/p&gt;

&lt;h2&gt;
  
  
  What must manufacturers actually do?
&lt;/h2&gt;

&lt;p&gt;Manufacturers must assess cybersecurity risk, meet the essential requirements in Annex I, document the result and keep handling vulnerabilities for a stated support period. In practice that turns into a set of engineering and governance habits rather than a single audit event.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Risk assessment:&lt;/strong&gt; Assess cybersecurity risk for the product and use it to decide how each essential requirement applies.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Secure design and defaults:&lt;/strong&gt; Build to the Annex I requirements throughout planning, design, development and maintenance.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Component due diligence:&lt;/strong&gt; Check third-party components, including open source, and identify them. &lt;a href="https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ%3AL_202402847" rel="noopener noreferrer"&gt;Recital 77&lt;/a&gt; expects a software bill of materials but says it need not be public. Our &lt;a href="https://getsecureslate.com/blog/eu-cyber-resilience-act-sbom" rel="noopener noreferrer"&gt;SBOM guide for the CRA&lt;/a&gt; covers that piece in depth.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Support period:&lt;/strong&gt; Tell buyers the end date, by month and year, at the time of purchase. &lt;a href="https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ%3AL_202402847" rel="noopener noreferrer"&gt;Article 13(8)&lt;/a&gt; requires a support period of at least five years, unless the product is expected to be in use for less time.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Technical documentation:&lt;/strong&gt; Keep it ready for market surveillance authorities.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Declaration and CE marking:&lt;/strong&gt; Draw up the EU declaration of conformity and affix the CE mark after a successful assessment.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Start by naming an owner. In our experience, CRA programs stall when product, security and legal each assume another team is leading.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which conformity assessment route applies?
&lt;/h2&gt;

&lt;p&gt;Your route depends on how the product is classified, and most products can self-assess. The &lt;a href="https://digital-strategy.ec.europa.eu/en/policies/cra-summary" rel="noopener noreferrer"&gt;Commission summary&lt;/a&gt; sets out the options, with product categories listed in Annexes III and IV and described further in Implementing Regulation (EU) 2025/2392.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Product class&lt;/th&gt;
&lt;th&gt;Route&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Default&lt;/td&gt;
&lt;td&gt;Self-assessment, a notified body, or a certification scheme where specified&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Important, class I&lt;/td&gt;
&lt;td&gt;Self-assessment only if harmonised standards, common specifications or a certification scheme were applied. Otherwise a notified body&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Important, class II&lt;/td&gt;
&lt;td&gt;Third-party assessment, or a certification scheme where available&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Critical&lt;/td&gt;
&lt;td&gt;Third-party assessment, or a certification scheme where available. No self-assessment&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Check your classification early. If you land in class I, the availability of harmonised standards decides whether you can self-assess, and the Commission's timeline expects standardisation work to continue through October 2027.&lt;/p&gt;

&lt;p&gt;How audit-ready are you? Answer 12 questions and get your readiness score&lt;/p&gt;

&lt;h2&gt;
  
  
  How does CRA incident and vulnerability reporting work?
&lt;/h2&gt;

&lt;p&gt;Manufacturers must report actively exploited vulnerabilities and severe incidents affecting product security on a fixed clock, through a single EU platform. &lt;a href="https://www.enisa.europa.eu/node/19439" rel="noopener noreferrer"&gt;ENISA's FAQ&lt;/a&gt; confirms that the Single Reporting Platform became operational on 11 September 2026, and that notifications reach the CSIRT for the manufacturer's main EU establishment.&lt;/p&gt;

&lt;p&gt;The clock runs from when you become aware of the issue:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Step&lt;/th&gt;
&lt;th&gt;Deadline&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Early warning&lt;/td&gt;
&lt;td&gt;Within 24 hours&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Notification with initial assessment&lt;/td&gt;
&lt;td&gt;Within 72 hours&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Final report, actively exploited vulnerability&lt;/td&gt;
&lt;td&gt;No later than 14 days after a fix or mitigation is available&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Final report, severe incident&lt;/td&gt;
&lt;td&gt;Within one month of the 72-hour notification&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;If you have not yet prepared, three actions close most of the gap:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;Create your organization's account on the platform before you need it. ENISA requires EU Login with multi-factor authentication.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Add a CRA decision point to your incident runbook, so that on-call staff know what triggers the 24-hour clock.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Inventory every in-scope product already on the EU market, because the duty covers them too.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  What happens if you do not comply?
&lt;/h2&gt;

&lt;p&gt;Member States enforce the CRA through national market surveillance authorities, and fines can reach the higher of EUR 15 million or 2.5% of worldwide annual turnover for breaches of the essential requirements or the Article 13 and 14 duties. Article 64 sets lower ceilings for other failures:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Breach&lt;/th&gt;
&lt;th&gt;Maximum fine&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Essential requirements, Articles 13 and 14&lt;/td&gt;
&lt;td&gt;EUR 15 million or 2.5% of worldwide annual turnover&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Other listed obligations&lt;/td&gt;
&lt;td&gt;EUR 10 million or 2% of worldwide annual turnover&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Incorrect or misleading information to authorities&lt;/td&gt;
&lt;td&gt;EUR 5 million or 1% of worldwide annual turnover&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Two carve-outs matter for smaller teams. Microenterprises and small enterprises cannot be fined for missing the 24-hour early warning deadline, and open-source software stewards are not subject to these fines. The commercial risk can be just as serious, since non-conforming products may lose access to the EU market.&lt;/p&gt;

&lt;h2&gt;
  
  
  How SecureSlate helps
&lt;/h2&gt;

&lt;p&gt;The CRA does not have a built-in framework page in SecureSlate today, so you would set it up as a custom framework with your own controls and map them to programs you already run, such as ISO 27001 or NIS 2. From there, &lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;SecureSlate&lt;/a&gt; supports the surrounding work:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Custom frameworks and controls to track CRA duties alongside SOC 2, ISO 27001 and NIS 2&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Code security and secrets detection to surface issues in your repositories&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Vendor risk management for third-party due diligence&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Risk management and audit management to keep assessments and evidence in one place&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;SecureSlate does not perform CRA conformity assessments, issue declarations or file reports on your behalf.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;Start a SecureSlate trial&lt;/a&gt; to set up your CRA control set.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Does the Cyber Resilience Act apply to SaaS?
&lt;/h3&gt;

&lt;p&gt;Usually not on its own. Recital 12 notes that NIS 2 applies to SaaS, and a cloud service only falls under the CRA when it is remote data processing that a product needs to perform a function.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is open-source software covered by the CRA?
&lt;/h3&gt;

&lt;p&gt;Free and open-source software is outside the rules when not supplied in a commercial activity. Open-source software stewards have lighter duties under Article 24, and their reporting duties begin on 11 December 2027.&lt;/p&gt;

&lt;h3&gt;
  
  
  Do I need to report vulnerabilities in products sold before 2027?
&lt;/h3&gt;

&lt;p&gt;Yes. The reporting obligations have applied since 11 September 2026, including to products that were placed on the market before 11 December 2027.&lt;/p&gt;

&lt;h3&gt;
  
  
  How long must a product be supported under the CRA?
&lt;/h3&gt;

&lt;p&gt;Article 13(8) requires a support period of at least five years, unless the product is expected to be in use for a shorter time. Manufacturers must state the end date clearly at purchase.&lt;/p&gt;

&lt;h2&gt;
  
  
  Disclaimer (legal note)
&lt;/h2&gt;

&lt;p&gt;This article is for general information only and is not legal, regulatory or professional advice. Requirements vary by framework, industry and jurisdiction. Consult qualified advisors for your specific obligations.&lt;/p&gt;

&lt;h3&gt;
  
  
  Need compliance without the complexity?
&lt;/h3&gt;

&lt;p&gt;SecureSlate automates ISO 27001, SOC 2, GDPR, HIPAA, and more. Built for growing teams. See it in action.&lt;/p&gt;

&lt;p&gt;Learn more: &lt;a href="https://getsecureslate.com/products/multi-framework" rel="noopener noreferrer"&gt;Compliance automation with SecureSlate&lt;/a&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://getsecureslate.com/blog/eu-cyber-resilience-act" rel="noopener noreferrer"&gt;SecureSlate blog&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>security</category>
      <category>webdev</category>
      <category>devops</category>
      <category>opensource</category>
    </item>
    <item>
      <title>AI Startup Enterprise Security Requirements: What Buyers Will Ask For</title>
      <dc:creator>SecureSlate</dc:creator>
      <pubDate>Mon, 16 Nov 2026 14:01:38 +0000</pubDate>
      <link>https://dev.to/secureslate/ai-startup-enterprise-security-requirements-what-buyers-will-ask-for-b61</link>
      <guid>https://dev.to/secureslate/ai-startup-enterprise-security-requirements-what-buyers-will-ask-for-b61</guid>
      <description>&lt;p&gt;&lt;strong&gt;Short answer:&lt;/strong&gt; Enterprises expect AI startups to meet normal SaaS security requirements, usually shown with a SOC 2 report, and then to explain how the AI itself is governed. That means clear commitments on customer data and model training, disclosed model providers, controls for prompt injection and agent permissions, and documented human oversight.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Related guides:&lt;/strong&gt; &lt;a href="https://getsecureslate.com/blog/soc-2-for-ai-startups-what-regulators-watch-and-how-to-stay-compliant" rel="noopener noreferrer"&gt;SOC 2 for AI startups&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/enterprise-security-review" rel="noopener noreferrer"&gt;enterprise security review&lt;/a&gt;, &lt;a href="https://getsecureslate.com/blog/the-ai-vendor-questionnaire-that-exposes-hidden-security-risks" rel="noopener noreferrer"&gt;the AI vendor questionnaire&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Key takeaways
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;A strong AI demo rarely shortens security review. In our experience it often adds questions, because the product touches more sensitive data.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Write your data use position first, since it shapes the answers to many of the other questions.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Treat each model provider as a subprocessor and disclose it.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Expect reviewers to ask about prompt injection, sensitive data exposure and what your agents are allowed to do.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;SOC 2 is usually the entry ticket. ISO 42001 can help when buyers ask how AI is governed, but it is not a replacement.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why do enterprises review AI startups differently?
&lt;/h2&gt;

&lt;p&gt;Enterprises review AI startups differently because an AI product adds new data paths, third-party model dependencies and non-deterministic behavior on top of ordinary application risk. A standard questionnaire was not written for a system that can be steered through its inputs or take actions on its own.&lt;/p&gt;

&lt;p&gt;Buyers are also working from published guidance now. The &lt;a href="https://www.nist.gov/itl/ai-risk-management-framework" rel="noopener noreferrer"&gt;NIST AI Risk Management Framework&lt;/a&gt; and its &lt;a href="https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence" rel="noopener noreferrer"&gt;Generative AI Profile, NIST AI 600-1&lt;/a&gt;, published in July 2024, give risk teams a shared vocabulary, and many AI questionnaires borrow from them. Security engineers often reach for the &lt;a href="https://genai.owasp.org/llm-top-10/" rel="noopener noreferrer"&gt;OWASP Top 10 for LLM Applications&lt;/a&gt; as well.&lt;/p&gt;

&lt;p&gt;The people in the room change too. In our experience, a deal with an AI product can pull in security, legal, privacy and an AI governance committee, each with a separate set of concerns. Preparing one reusable evidence pack keeps those parallel reviews from consuming your engineers.&lt;/p&gt;

&lt;h2&gt;
  
  
  What evidence will an enterprise security team request?
&lt;/h2&gt;

&lt;p&gt;Expect requests in two layers: the baseline any SaaS vendor must show, and a set of AI-specific materials. If you have the first layer and not the second, you will still get stuck.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Layer&lt;/th&gt;
&lt;th&gt;What to have ready&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;SaaS baseline&lt;/td&gt;
&lt;td&gt;SOC 2 report or a dated plan to obtain one, penetration test summary, security policies, incident response plan, subprocessor list, data processing agreement&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI data handling&lt;/td&gt;
&lt;td&gt;Written data use and training policy, retention for prompts and outputs, tenant isolation description&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI supply chain&lt;/td&gt;
&lt;td&gt;List of model providers and hosting locations, provider terms that support your commitments&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI behavior and safety&lt;/td&gt;
&lt;td&gt;Threat model for AI features, evaluation approach, known limitations, human oversight description&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI governance&lt;/td&gt;
&lt;td&gt;AI policy, named owner, process for approving model changes and new AI use cases&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Keep this material in one place and gate access through an NDA where needed. A &lt;a href="https://getsecureslate.com/blog/build-a-high-conversion-trust-center-in-5-steps-to-prove-security-fast" rel="noopener noreferrer"&gt;trust center&lt;/a&gt; lets buyers help themselves and cuts back-and-forth.&lt;/p&gt;

&lt;p&gt;See SecureSlate map one control across SOC 2, ISO 27001 and HIPAA&lt;/p&gt;

&lt;h2&gt;
  
  
  How should you answer questions about customer data and training?
&lt;/h2&gt;

&lt;p&gt;Give a direct, written answer that says whether customer data trains any model, who can see it, how long it is kept and where it is processed. Vague language is the most common reason this part of a review drags on, in our experience.&lt;/p&gt;

&lt;p&gt;Your answer should address each of these points:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Training:&lt;/strong&gt; Do you or your model providers train on customer content? If the answer differs by plan or setting, say so.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Retention:&lt;/strong&gt; How long are prompts, outputs, embeddings and logs kept, and can a customer change this?&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Isolation:&lt;/strong&gt; How are vector stores, caches and fine-tuned models kept separate between tenants?&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Provider terms:&lt;/strong&gt; Does each provider's contract actually support what you have promised? Your commitment is only as strong as the agreement behind it.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Human access:&lt;/strong&gt; Which employees can view customer content, under what approval and with what logging?&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Then make sure the product, contract and security documentation all say the same thing. Reviewers compare them.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which AI-specific controls do reviewers test?
&lt;/h2&gt;

&lt;p&gt;Reviewers focus on how the system handles hostile input, what it can leak and how far an agent can act. The 2025 OWASP list is a practical checklist, because it names prompt injection as LLM01, sensitive information disclosure as LLM02 and excessive agency as LLM06.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Risk&lt;/th&gt;
&lt;th&gt;What a reviewer wants to see&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Prompt injection&lt;/td&gt;
&lt;td&gt;Untrusted content is separated from instructions, and tests cover indirect injection through documents and web content&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Sensitive information disclosure&lt;/td&gt;
&lt;td&gt;Retrieval respects the requesting user's permissions, and outputs are checked for leaked data&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Supply chain&lt;/td&gt;
&lt;td&gt;Models and libraries are inventoried, versioned and approved&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Excessive agency&lt;/td&gt;
&lt;td&gt;Agents use least-privilege credentials, and high-impact actions need human approval&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Improper output handling&lt;/td&gt;
&lt;td&gt;Model output is treated as untrusted before it reaches code, queries or browsers&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Logging ties these together. Be ready to show that you can reconstruct what a user asked, what the model returned and which tools it called, within limits your retention policy allows. Be candid about residual risk too, since no current defense fully removes prompt injection. A reviewer will trust "here is how we limit the damage" more than "we have solved it."&lt;/p&gt;

&lt;h2&gt;
  
  
  Do you need SOC 2, ISO 42001 or both?
&lt;/h2&gt;

&lt;p&gt;Most AI startups selling to enterprises in the US should start with SOC 2, then add ISO 42001 if buyers begin asking for evidence of AI governance. They answer different questions: SOC 2 reports on controls for your service, while &lt;a href="https://www.iso.org/standard/42001" rel="noopener noreferrer"&gt;ISO/IEC 42001&lt;/a&gt; specifies a management system for how an organization governs AI.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Situation&lt;/th&gt;
&lt;th&gt;Sensible order&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;US mid-market or enterprise buyers&lt;/td&gt;
&lt;td&gt;SOC 2 first, AI governance documents alongside it&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Buyers with formal AI committees&lt;/td&gt;
&lt;td&gt;SOC 2, then ISO 42001&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Europe-heavy pipeline&lt;/td&gt;
&lt;td&gt;ISO 27001 or SOC 2 based on buyer preference, then ISO 42001&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Healthcare customers&lt;/td&gt;
&lt;td&gt;SOC 2 and HIPAA readiness before any AI-specific certification&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Neither one proves that your model is accurate or unbiased. For that, buyers will look at your evaluations. Our &lt;a href="https://getsecureslate.com/blog/iso-42001-framework-guide" rel="noopener noreferrer"&gt;ISO 42001 framework guide&lt;/a&gt; goes deeper on the certification path.&lt;/p&gt;

&lt;p&gt;How audit-ready are you? Answer 12 questions and get your readiness score&lt;/p&gt;

&lt;h2&gt;
  
  
  What changes when you sell to healthcare or EU buyers?
&lt;/h2&gt;

&lt;p&gt;Regulated buyers add legal requirements on top of security review. Two situations come up most often for AI startups.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Healthcare.&lt;/strong&gt; If your product handles protected health information for a covered entity or business associate, you are likely a business associate. HHS guidance says that &lt;a href="https://www.hhs.gov/hipaa/for-professionals/covered-entities/sample-business-associate-agreement-provisions/index.html" rel="noopener noreferrer"&gt;business associates need written contracts with subcontractors&lt;/a&gt; that create, receive, maintain or transmit that information on their behalf. A model provider that processes PHI for you falls into that chain, so confirm a BAA is available before you route clinical data to it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;European Union.&lt;/strong&gt; Expect buyers to ask for your role under the &lt;a href="https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai" rel="noopener noreferrer"&gt;EU AI Act&lt;/a&gt;, usually whether you are a provider or a deployer, and how your system is classified. Application dates for some obligations have been the subject of amending legislation, so check the current position before you put a date in a contract or questionnaire answer. Under GDPR, your model providers are also likely to be &lt;a href="https://getsecureslate.com/blog/gdpr-sub-processors" rel="noopener noreferrer"&gt;sub-processors&lt;/a&gt; that you must disclose.&lt;/p&gt;

&lt;h2&gt;
  
  
  What should an early-stage team do first?
&lt;/h2&gt;

&lt;p&gt;Build the evidence in the order buyers will ask for it, and keep the first version short. A two-page document that is accurate beats a long one that is out of date.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Write it down:&lt;/strong&gt; data use policy, model provider list and a short AI feature inventory.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Close the obvious gaps:&lt;/strong&gt; tenant isolation, least-privilege agent credentials and logging.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Get independent assurance:&lt;/strong&gt; start SOC 2 and commission a penetration test that includes AI features.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;&lt;strong&gt;Package and reuse:&lt;/strong&gt; keep an evidence pack and a questionnaire answer library, and update both whenever you change a model or provider.&lt;/p&gt;&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Assign one owner for keeping this current. Model changes are frequent, and an out-of-date subprocessor list can undermine a review that was otherwise going well.&lt;/p&gt;

&lt;h2&gt;
  
  
  How SecureSlate helps
&lt;/h2&gt;

&lt;p&gt;SecureSlate gives small teams one place to run this work. Multi-framework control mapping covers SOC 2, ISO 27001, ISO 42001, HIPAA, GDPR and the EU AI Act, so shared controls do not have to be tested twice, and custom controls can capture your AI-specific commitments.&lt;/p&gt;

&lt;p&gt;Agentless read-only cloud integrations and access reviews support evidence collection, code security and secrets detection cover your repositories, and vendor risk management helps you track model providers and other suppliers. A trust center and security questionnaire automation help you reuse approved answers during a review. Outcomes still depend on your controls, scope and auditor.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://app.getsecureslate.com/trial/register" rel="noopener noreferrer"&gt;Start a SecureSlate trial&lt;/a&gt; to build your enterprise evidence pack.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Do AI startups need SOC 2 before selling to enterprises?
&lt;/h3&gt;

&lt;p&gt;Not legally, but many enterprise buyers expect it. If you do not have a report yet, share a dated plan, a current penetration test summary and your policies so that the review can proceed.&lt;/p&gt;

&lt;h3&gt;
  
  
  Should we list our model provider as a subprocessor?
&lt;/h3&gt;

&lt;p&gt;Yes, if it processes customer data on your behalf. Include its role, the type of data involved and where processing takes place, and tell customers when that changes.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is ISO 42001 required to sell AI to enterprises?
&lt;/h3&gt;

&lt;p&gt;It is a voluntary standard, and not every buyer will ask for it. It is most useful when customers want independent evidence of how you govern AI.&lt;/p&gt;

&lt;h3&gt;
  
  
  How do we answer a questionnaire not written for AI products?
&lt;/h3&gt;

&lt;p&gt;Answer the questions as asked, then attach a short AI addendum covering data use, model providers, AI-specific controls and human oversight. That often heads off a second round of questions.&lt;/p&gt;

&lt;h2&gt;
  
  
  Disclaimer (legal note)
&lt;/h2&gt;

&lt;p&gt;This article is for general information only and is not legal, regulatory or professional advice. Requirements vary by framework, industry and jurisdiction. Consult qualified advisors for your specific obligations.&lt;/p&gt;

&lt;h3&gt;
  
  
  Need compliance without the complexity?
&lt;/h3&gt;

&lt;p&gt;SecureSlate automates ISO 27001, SOC 2, GDPR, HIPAA, and more. Built for growing teams. See it in action.&lt;/p&gt;

&lt;p&gt;Find compliance gaps in 30 seconds&lt;/p&gt;

&lt;p&gt;Filed under: &lt;a href="https://getsecureslate.com/blog/topic/ai" rel="noopener noreferrer"&gt;AI&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Learn more: &lt;a href="https://getsecureslate.com/frameworks/iso-42001" rel="noopener noreferrer"&gt;ISO 42001 compliance with SecureSlate&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Author:  SecureSlate Team&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Originally published on the &lt;a href="https://getsecureslate.com/blog/ai-startup-enterprise-security-requirements" rel="noopener noreferrer"&gt;SecureSlate blog&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>security</category>
      <category>ai</category>
      <category>startup</category>
      <category>compliance</category>
    </item>
  </channel>
</rss>
