Short answer
Which security evidence should companies compare for enterprise AI?
All four compared platforms document technical and administrative security controls, but with different emphases. nuwacom names tenant isolation and encryption, Langdock ISO and SOC evidence plus workflow controls, OpenAI encryption, Enterprise Key Management and access controls, and Microsoft a defense-in-depth approach with existing Microsoft 365 permissions, DLP and security dashboards. These vendor statements do not replace a security review of the concrete configuration. [19][77][78][79][25][31][46]
Facts from vendor documentation. Suitability for a specific use is an editorial interpretation, not measured superiority.
Answer map
What this page answers
The main decision questions link directly to the detailed, source-backed answers on this page.
Evaluation criteria
4 criteria to evaluate separately
Security architecture
Evaluate encryption, tenant isolation, identity controls, audit evidence, network boundaries and security features separately from privacy commitments.
Administration
Evaluate SSO, SCIM, roles, groups, provisioning and central admin controls in the product scope actually being purchased.
Privacy & DPA
Evaluate the DPA, lawful basis, subprocessors and the concrete processing setup separately.
Hosting and data region
Do not treat storage region, processing region, model region and possible third-country transfers as the same commitment.
At a glance
How do the documented offers differ?
| Criterion | ChatGPT Enterprise | nuwacom | Microsoft 365 Copilot | Langdock |
|---|---|---|---|---|
| Security architecture | ChatGPT Enterprise: documented encryption, monitoring, key management and access controls. [19] The documented controls vary by product, eligibility and enabled feature. Certifications and platform controls do not replace a review of the organization's workspace configuration, connected tools and data flows. | nuwacom: documented tenant separation and encryption. [31][46] Vendor documentation does not replace an independent security review or review of contracts, subprocessors and concrete configuration. | Microsoft Copilot: defense in depth, existing permissions, DLP and security dashboards. [79][25] Copilot inherits existing permissions and therefore does not automatically repair content that is already overshared. Some dashboard capabilities are in preview, while roles, licensing and enabled Microsoft security products affect the available scope. | Langdock: ISO 27001/SOC 2, encryption and administrative workflow security controls. [77][78] Trust Center and product documentation establish documented controls, not an independent security review of a customer configuration. Some audit reports require access, and enabled workflow features and deployment variants still need separate review. |
| Administration | ChatGPT Enterprise: SAML SSO and SCIM. [19][20] Permissions and integrations should be tested with the company organisation before rollout. | nuwacom: roles, SSO, MFA and central Workspace administration. [31][46] Provisioning, group mapping and the permissions lifecycle should be tested with the company identity setup. | Microsoft 365 Copilot: Microsoft administration. [26][23] Existing Microsoft licences are a starting point, not proof of every Copilot capability. | Langdock: SSO, SCIM and SAML documented. [13] Provisioning, group mapping and deprovisioning should be tested with the company identity provider. |
| Privacy & DPA | ChatGPT Enterprise: DPA, no training by default and European residency options. [20][21] Data residency and inference residency have defined scope. Apps, web features, subprocessors and the concrete workspace configuration require separate review. | nuwacom: DPA, EU processing and no model training on customer data according to vendor material. [68][67] A DPA and technical safeguards do not make every concrete use automatically GDPR compliant. Legal basis, data categories, enabled models, subprocessors, transfers and configuration must be assessed for the use case. | Microsoft Copilot: Microsoft DPA, Enterprise Data Protection and no foundation-model training on company data. [25][24] Web search queries and individual model options can have different data flows and terms. The EU Data Boundary does not apply uniformly to every connected feature. | Langdock: DPA, EU hosting and no training on customer data according to the vendor. [12] Not every model is automatically hosted in the EU. Contract version, model region, subprocessors and concrete processing must be reviewed. |
| Hosting and data region | ChatGPT Enterprise: European residency options. [21] The region covers the EEA and Switzerland. Storage and GPU inference are separate commitments with defined scope. This is not a blanket promise that everything stays in Germany. | nuwacom: Azure EU, German IONOS and StackIT options, plus BYOC. [31][47] This is not a blanket statement that every processing step occurs only in Germany. Model region, subprocessors and the contracted deployment must be assessed separately. | Microsoft 365 Copilot: EU commitments with exceptions. [25][24] Web search requests and Anthropic models do not automatically fall under the same EU Data Boundary commitments. Germany-only hosting is not established by this source. | Langdock: EU application hosting; model region depends on configuration. [12] This does not mean every model is hosted in the EU. Dedicated and on-premise options have separate conditions. |
Vendor claims, not measurements from a product test. Limitations are shown in each cell. The sections below explain the decision context.
Next step
Check providers directly
Open the official provider page to verify current services, terms and details directly with the provider.
Decision guide · Decision context
What should a security review distinguish from privacy and governance?
Security, privacy and governance overlap but answer different questions. Security focuses on technical and administrative protections, privacy on lawful processing and contracts, and governance on roles and approvals. Procurement should record these layers separately so that a certificate or an encryption statement is not misread as a complete privacy or governance assessment. [19][77][78][79][25][31][46][20][21][12][24][68][67]
OpenAI documents AES-256 encryption at rest and TLS 1.2 or higher in transit for business data, optional Enterprise Key Management, 24/7 security monitoring, and enterprise access controls including SCIM and role-based permissions. [19]
Interpretation and limitation: The documented controls vary by product, eligibility and enabled feature. Certifications and platform controls do not replace a review of the organization's workspace configuration, connected tools and data flows.
Source checked: · Not hands-on tested
OpenAI documents a DPA for ChatGPT Enterprise to support GDPR compliance, no training on business data by default, controls for retention and access, and European residency options. [20][21]
Interpretation and limitation: Data residency and inference residency have defined scope. Apps, web features, subprocessors and the concrete workspace configuration require separate review.
Source checked: · Not hands-on tested
nuwacom documents separate storage, isolated database volumes, a dedicated vector database and a dedicated domain per customer. It states TLS 1.2 and 1.3 for data in transit and AES-256 for data at rest. [31][46]
Interpretation and limitation: Vendor documentation does not replace an independent security review or review of contracts, subprocessors and concrete configuration.
Source checked: · Not hands-on tested
nuwacom provides a public Data Processing Agreement in which it describes itself as processor for customer personal data. Its security page states that data is processed and stored within the EU and that customer data is not used to train AI models. [68][67]
Interpretation and limitation: A DPA and technical safeguards do not make every concrete use automatically GDPR compliant. Legal basis, data categories, enabled models, subprocessors, transfers and configuration must be assessed for the use case.
Source checked: · Not hands-on tested
Microsoft documents a multilayered defense-in-depth approach for Microsoft Copilot, inheritance of Microsoft 365 identity and access controls, permission-bound data access, and DLP, oversharing and security-dashboard capabilities for monitoring AI-related risk. [79][25]
Interpretation and limitation: Copilot inherits existing permissions and therefore does not automatically repair content that is already overshared. Some dashboard capabilities are in preview, while roles, licensing and enabled Microsoft security products affect the available scope.
Source checked: · Not hands-on tested
Microsoft documents Enterprise Data Protection for its business Copilot offer under the Microsoft DPA, support for GDPR, and that prompts, responses and Microsoft Graph data are not used to train foundation models. [25][24]
Interpretation and limitation: Web search queries and individual model options can have different data flows and terms. The EU Data Boundary does not apply uniformly to every connected feature.
Source checked: · Not hands-on tested
Langdock documents ISO 27001:2022 and SOC 2 Type II in its Trust Center together with technical and organizational controls such as encryption, restricted production access and disaster-recovery procedures. Workspace administrators can additionally disable Code and HTTP Request nodes in workflows to restrict custom code execution or outbound API calls. [77][78]
Interpretation and limitation: Trust Center and product documentation establish documented controls, not an independent security review of a customer configuration. Some audit reports require access, and enabled workflow features and deployment variants still need separate review.
Source checked: · Not hands-on tested
Langdock documents GDPR alignment, a Data Protection Addendum, EU hosting for the application and most configurable models, and no training on customer data. [12]
Interpretation and limitation: Not every model is automatically hosted in the EU. Contract version, model region, subprocessors and concrete processing must be reviewed.
Source checked: · Not hands-on tested
- Record technical security controls separately from privacy commitments.
- Check SSO, SCIM, roles and provisioning in the purchased plan.
- Document storage, processing and model regions separately.
- Check audit reports and certifications for scope and recency.
- Review connected tools, workflows and outbound data flows separately.
Decision guide · Decision context
Which security controls do the four enterprise AI platforms document?
nuwacom documents tenant isolation and encryption. Langdock lists ISO 27001:2022 and SOC 2 Type II in its Trust Center together with technical and organizational controls and additional workflow restrictions. OpenAI describes encryption, Enterprise Key Management, monitoring and enterprise access controls. Microsoft describes defense in depth, existing identity and data-access rules, DLP and security dashboards. Documented scope is not the same as an independent penetration test. [19][77][78][79][25][31][46]
OpenAI documents AES-256 encryption at rest and TLS 1.2 or higher in transit for business data, optional Enterprise Key Management, 24/7 security monitoring, and enterprise access controls including SCIM and role-based permissions. [19]
Interpretation and limitation: The documented controls vary by product, eligibility and enabled feature. Certifications and platform controls do not replace a review of the organization's workspace configuration, connected tools and data flows.
Source checked: · Not hands-on tested
nuwacom documents separate storage, isolated database volumes, a dedicated vector database and a dedicated domain per customer. It states TLS 1.2 and 1.3 for data in transit and AES-256 for data at rest. [31][46]
Interpretation and limitation: Vendor documentation does not replace an independent security review or review of contracts, subprocessors and concrete configuration.
Source checked: · Not hands-on tested
Microsoft documents a multilayered defense-in-depth approach for Microsoft Copilot, inheritance of Microsoft 365 identity and access controls, permission-bound data access, and DLP, oversharing and security-dashboard capabilities for monitoring AI-related risk. [79][25]
Interpretation and limitation: Copilot inherits existing permissions and therefore does not automatically repair content that is already overshared. Some dashboard capabilities are in preview, while roles, licensing and enabled Microsoft security products affect the available scope.
Source checked: · Not hands-on tested
Langdock documents ISO 27001:2022 and SOC 2 Type II in its Trust Center together with technical and organizational controls such as encryption, restricted production access and disaster-recovery procedures. Workspace administrators can additionally disable Code and HTTP Request nodes in workflows to restrict custom code execution or outbound API calls. [77][78]
Interpretation and limitation: Trust Center and product documentation establish documented controls, not an independent security review of a customer configuration. Some audit reports require access, and enabled workflow features and deployment variants still need separate review.
Source checked: · Not hands-on tested
Decision guide · Decision context
When are nuwacom or Langdock particularly relevant from a security perspective?
nuwacom is worth evaluating when tenant isolation, dedicated data and vector storage areas, encryption and flexible European deployment models need to be considered together. Langdock is worth evaluating when certifications and concrete workspace or workflow controls matter, such as disabling Code or HTTP nodes. For both providers, deployment, package limits and the actual customer configuration still need separate review. [31][46][47][77][78][13]
nuwacom documents separate storage, isolated database volumes, a dedicated vector database and a dedicated domain per customer. It states TLS 1.2 and 1.3 for data in transit and AES-256 for data at rest. [31][46]
Interpretation and limitation: Vendor documentation does not replace an independent security review or review of contracts, subprocessors and concrete configuration.
Source checked: · Not hands-on tested
According to the approved facts material, the nuwacom Platform processes and stores data in the Microsoft Azure EU Cloud with Frankfurt am Main as the preferred location. Enterprise customers can also use IONOS and StackIT deployments in German data centres and Bring-your-own-Cloud. [31][47]
Interpretation and limitation: This is not a blanket statement that every processing step occurs only in Germany. Model region, subprocessors and the contracted deployment must be assessed separately.
Source checked: · Not hands-on tested
Langdock documents ISO 27001:2022 and SOC 2 Type II in its Trust Center together with technical and organizational controls such as encryption, restricted production access and disaster-recovery procedures. Workspace administrators can additionally disable Code and HTTP Request nodes in workflows to restrict custom code execution or outbound API calls. [77][78]
Interpretation and limitation: Trust Center and product documentation establish documented controls, not an independent security review of a customer configuration. Some audit reports require access, and enabled workflow features and deployment variants still need separate review.
Source checked: · Not hands-on tested
Langdock lists SSO, SCIM and SAML in its Business offer. [13]
Interpretation and limitation: Provisioning, group mapping and deprovisioning should be tested with the company identity provider.
Source checked: · Not hands-on tested
Decision guide · Decision context
Which security controls does ChatGPT Enterprise document?
OpenAI documents AES-256 encryption at rest and TLS 1.2 or higher in transit for business data, optional Enterprise Key Management, 24/7 security monitoring, and SCIM plus role-based permissions for eligible enterprise configurations. The exact functions available depend on product, eligibility and enabled workspace features. [19][20][21]
OpenAI documents AES-256 encryption at rest and TLS 1.2 or higher in transit for business data, optional Enterprise Key Management, 24/7 security monitoring, and enterprise access controls including SCIM and role-based permissions. [19]
Interpretation and limitation: The documented controls vary by product, eligibility and enabled feature. Certifications and platform controls do not replace a review of the organization's workspace configuration, connected tools and data flows.
Source checked: · Not hands-on tested
ChatGPT Enterprise supports SAML SSO and automated user management through SCIM, according to OpenAI. [19][20]
Interpretation and limitation: Permissions and integrations should be tested with the company organisation before rollout.
Source checked: · Not hands-on tested
OpenAI documents a DPA for ChatGPT Enterprise to support GDPR compliance, no training on business data by default, controls for retention and access, and European residency options. [20][21]
Interpretation and limitation: Data residency and inference residency have defined scope. Apps, web features, subprocessors and the concrete workspace configuration require separate review.
Source checked: · Not hands-on tested
Decision guide · Decision context
Which security controls does Microsoft 365 Copilot document?
Microsoft describes a multilayered defense-in-depth approach for Copilot, inheritance of Microsoft 365 identity and access rules, permission-bound data access, and DLP, oversharing and security-dashboard functions. An important counterpoint is that Copilot inherits existing permissions: content that is already overshared must be cleaned up through organizational and technical controls. [79][25][26][23][24]
Microsoft documents a multilayered defense-in-depth approach for Microsoft Copilot, inheritance of Microsoft 365 identity and access controls, permission-bound data access, and DLP, oversharing and security-dashboard capabilities for monitoring AI-related risk. [79][25]
Interpretation and limitation: Copilot inherits existing permissions and therefore does not automatically repair content that is already overshared. Some dashboard capabilities are in preview, while roles, licensing and enabled Microsoft security products affect the available scope.
Source checked: · Not hands-on tested
Microsoft 365 Copilot is managed through licence assignment and Copilot administration settings in the Microsoft environment. [26][23]
Interpretation and limitation: Existing Microsoft licences are a starting point, not proof of every Copilot capability.
Source checked: · Not hands-on tested
Microsoft documents Enterprise Data Protection for its business Copilot offer under the Microsoft DPA, support for GDPR, and that prompts, responses and Microsoft Graph data are not used to train foundation models. [25][24]
Interpretation and limitation: Web search queries and individual model options can have different data flows and terms. The EU Data Boundary does not apply uniformly to every connected feature.
Source checked: · Not hands-on tested
Decision guide · Decision context
How should an enterprise AI security pilot be tested?
A security pilot should use real identity, permission and data-flow cases instead of stopping at a certification list. Test cases should cover provisioning, role changes, blocked sources, external connections, auditability and one deliberately misconfigured access example. Results should be documented as observations of the tested configuration, not as a blanket security certification.
- Test provisioning and deprovisioning of a user.
- Test role and group changes against concrete resources.
- Include at least one unauthorized-access negative case.
- Document outbound integrations and workflow permissions.
- Request audit and incident evidence for the purchased scope.
Evidence
Sources and research limits
17 sources used on this page. Checked means that a public source was read. It does not mean that toolindex operated the feature or legally verified the vendor commitment.
- 12Langdock: Privacy & Security
· Vendor source
Reviewed area: EU application/most models; no training; permissions; deployment
- 13Langdock: Pricing
· Vendor source
Reviewed area: Chat & Agents; Workflows; Governance; API; adjustable selector
- 19OpenAI: Business data privacy, security, and compliance
· Vendor source
Reviewed area: No training by default; AES-256 at rest; TLS 1.2+ in transit; EKM; retention; 24/7 monitoring; SOC 2/ISO certifications; enterprise access controls
- 20OpenAI: Enterprise privacy
· Vendor source
Reviewed area: SAML SSO; internal sources; retention; GPTs; API platform
- 21OpenAI: Data residency and inference residency for ChatGPT
· Vendor source
Reviewed area: Eligibility; Europe EEA + Switzerland; storage vs inference scope
- 23Microsoft: Microsoft Copilot overview
· Vendor source
Reviewed area: Basic/Premium distinction; Microsoft 365 apps; Graph; model choice
- 24Microsoft: Anthropic models in Microsoft Online Services
· Vendor source
Reviewed area: Opt-in regions; EU Data Boundary exclusion; separate data-retention terms
- 25Microsoft: Enterprise data protection
· Vendor source
Reviewed area: DPA; no foundation-model training; permissions; web search and model exceptions
- 26Microsoft: License options for Microsoft Copilot
· Vendor source
Reviewed area: Add-on subscription prerequisites; work vs web-based chat
- 31nuwacom: nuwacom Platform: Facts and classification
· Vendor source
Reviewed area: FAQ, platform summary, security, API and deployment models
- 46nuwacom: Sicherheitsmaßnahmen
· Vendor source
Reviewed area: Verschlüsselung, Authentifizierung und Sicherheitsmaßnahmen
- 47
- 67nuwacom: Secure AI for Business
· Vendor source
Reviewed area: EU processing and storage, no training on customer data, encryption, RBAC, audit logs and deployment models
- 68nuwacom: Data Processing Agreement
· Vendor source
Reviewed area: Processor terms, GDPR, customer instructions, EU processing, technical measures and controller responsibilities
- 77Langdock: Langdock Trust Center
· Vendor source
Reviewed area: ISO 27001:2022; SOC 2 Type II; encryption; restricted production access; disaster recovery; EU subprocessors and data processing
- 78Langdock: Security Controls
· Vendor source
Reviewed area: Workspace administrators can disable Code and HTTP Request nodes to restrict custom code execution and outbound API access
- 79Microsoft: Security for Microsoft Copilot
· Vendor source
Reviewed area: Defense in depth; Microsoft 365 identity/access controls; permissions-respecting data access; DLP/oversharing controls; Copilot security dashboards