SDU eScience Regulatory Compliance Annex
Version 1.6
Last modified: 15 September 2026
This Regulatory Compliance Annex (the “Annex”) is incorporated into the agreement under which SDU eScience has agreed to provide its services (as described at https://legal.cloud.sdu.dk/terms) and related technical support to Customer (the “Agreement”).
Capitalized terms defined in the Agreement, including in the Terms of Service and in the Data Processing and Security Terms (the “Data Processing Terms”), apply to this Annex.
0. Scope and Precedence
0.1 Purpose. This Annex records the obligations of the parties under the regulatory instruments listed in Section 0.2, and allocates responsibility between SDU eScience and Customer where those instruments require an allocation to be agreed.
0.2 Instruments Covered. This Annex addresses:
(a) Regulation (EU) 2016/679 (the “GDPR”);
(b) the Danish Act on supplementary provisions to the GDPR, lov nr. 502 af 23. maj 2018 om supplerende bestemmelser til forordningen om beskyttelse af fysiske personer i forbindelse med behandling af personoplysninger og om fri udveksling af sådanne oplysninger (“Databeskyttelsesloven”);
(c) Directive (EU) 2022/2555 (“NIS2”) and the Danish act implementing it, lov nr. 434 af 6. maj 2025 om foranstaltninger til sikring af et højt cybersikkerhedsniveau (NIS 2-loven), in force from 1 July 2025 (the “Danish NIS2 Implementation”);
(d) Regulation (EU) 2024/1689 (the “AI Act”).
0.3 Relationship to the Data Processing Terms. The Data Processing Terms remain the parties’ agreement under Article 28(3) of the GDPR. This Annex supplements them and does not replace them. To the extent of any conflict between this Annex and the Data Processing Terms on a matter of data protection, the Data Processing Terms prevail. To the extent of any conflict between this Annex and the remainder of the Agreement, this Annex prevails.
0.4 No Expansion of Liability. Nothing in this Annex expands the liability of either party beyond the limits set out in the Terms of Service.
0.5 Changes in Law. Where an instrument listed in Section 0.2 becomes applicable, is amended, or is supplemented by national implementing law after the Terms Effective Date, the parties will review this Annex and, acting reasonably, agree any amendments required under Section 14 (Amendments and Assignments) of the Data Processing Terms.
0.6 No Independent Processing. SDU eScience acts only on the Instructions of the Customer or, where Customer is a processor, of the relevant controller, and otherwise only as necessary for the operation of the infrastructure on which the Services are provided or, where an AI Service is provided, as described in Section 5.2.4(c) of the Data Processing Terms, in each case as set out in Section 5.2.4 (Limitation on Processing) of the Data Processing Terms. Nothing in this Annex authorizes SDU eScience to process Customer Personal Data for any other purpose, or constitutes an Instruction. Where an obligation in this Annex would otherwise require SDU eScience to process or disclose Customer Personal Data, SDU eScience will do so only on the Instructions or where required by law in accordance with Section 5.2.5 (Processing Required by Law) of the Data Processing Terms.
0.7 Technical Infrastructure; Allocation of Responsibility. SDU eScience provides technical infrastructure. Customer determines what it runs on that infrastructure and for what purpose. Except where this Annex or the Data Processing Terms expressly provide otherwise, SDU eScience does not monitor, review or assess Customer’s use of the Services, or Customer’s compliance with the instruments listed in Section 0.2, and has no obligation to do so. This reflects Section 7.2.4 (No Assessment of Customer Data by SDU eScience) of the Data Processing Terms.
Where this Annex states that Customer will or will not do something, that is an obligation of Customer; no corresponding obligation on SDU eScience to monitor, verify or enforce it is implied. Where this Annex gives SDU eScience a right to request evidence or to suspend, that is a right and not a duty, and SDU eScience does not warrant that exercising it will establish Customer’s compliance. This Section does not reduce the obligations this Annex expressly places on SDU eScience, in particular the cybersecurity risk-management measures in Section 3.2. Where SDU eScience provides an AI Service, Section 4.8 states the respects in which this Section is qualified for that Service.
1. General Data Protection Regulation
1.1 Applicable Data Protection Law. For the purposes of the Agreement, the data protection law applicable to SDU eScience’s processing of Customer Personal Data is the GDPR as supplemented by Databeskyttelsesloven and other Danish law. SDU eScience does not rely on, and makes no commitments in respect of, data protection law in force outside the European Union.
1.2 Roles. SDU eScience acts as a processor of Customer Personal Data, and Customer acts as controller or as processor, in each case as set out in Section 5.1 of the Data Processing Terms. This Annex does not alter that allocation.
1.3 Establishment and Supervisory Authority. SDU eScience is part of the University of Southern Denmark (Syddansk Universitet), a Danish public authority established in Denmark. Its lead supervisory authority for the purposes of Article 56 of the GDPR is the Danish Data Protection Agency (Datatilsynet), Carl Jacobsens Vej 35, 2500 Valby, Denmark.
1.4 Data Protection Officer. The Data Protection Officer designated under Article 37 of the GDPR may be contacted at dpo@sdu.dk, as set out in Section 12.1 of the Data Processing Terms.
1.5 Records of Processing. SDU eScience maintains records of processing activities carried out on behalf of Customer in accordance with Article 30(2) of the GDPR, and will make those records available to Datatilsynet on request.
1.6 Special Categories of Data. Customer will notify SDU eScience before using the Services to process personal data falling within Article 9 or Article 10 of the GDPR. Notification under this Section is required so that SDU eScience can determine whether additional Security Measures apply, and does not transfer to SDU eScience any responsibility for establishing a lawful basis for that processing.
1.7 Personal Data Breach Notification. SDU eScience will notify Customer of a Data Incident in accordance with Section 7.2 of the Data Processing Terms, which sets the period within which that notification must be given. Where an incident is also reportable under Section 3 of this Annex, SDU eScience will meet both sets of deadlines; the deadlines run independently of one another.
2. Databeskyttelsesloven
2.1 Supplementary National Law. Databeskyttelsesloven supplements the GDPR in Denmark. Where the Agreement refers to European Law or to applicable data protection law, that reference includes Databeskyttelsesloven.
2.2 Processing for Scientific and Statistical Purposes. SDU eScience provides the Services principally for research and other scientific work. Customer acknowledges that:
(a) personal data, including data falling within Article 9 of the GDPR, processed on the Services solely for the purposes of scientific or statistical research may be processed under § 10(1) of Databeskyttelsesloven;
(b) under § 10(2) of Databeskyttelsesloven, data processed under § 10(1) may not subsequently be processed for any purpose other than scientific or statistical research; and
(c) under § 10(3) of Databeskyttelsesloven, disclosure of such data to a third party may require prior authorization from Datatilsynet, and obtaining any such authorization is Customer’s responsibility.
Customer will not use the Services to process data received under § 10(1) for any commercial, administrative or decision-making purpose concerning the data subject.
2.3 Civil Registration Numbers. The Danish civil registration number (CPR-nummer) is subject to § 11 of Databeskyttelsesloven. Customer will:
(a) not upload CPR numbers to the Services unless Customer has established a lawful basis under § 11 for doing so;
(b) where CPR numbers are processed, apply pseudonymization or equivalent measures before processing unless the research design demonstrably requires the identifier; and
(c) not use the Services to publish or disclose CPR numbers.
2.4 Processing by Public Authorities. Where Customer is a Danish public authority, Customer remains responsible for compliance with §§ 5 to 9 of Databeskyttelsesloven in respect of its own processing, including any processing of data concerning criminal offences.
2.5 Retention and Public Archives. Customer acknowledges that SDU eScience, as a Danish public authority, may be subject to record-keeping and archiving obligations under Danish law, including arkivloven. Where such an obligation requires SDU eScience to retain Customer Data that would otherwise be deleted under Section 6 of the Data Processing Terms, SDU eScience will notify Customer, will retain only the data the obligation requires, and will continue to protect it under the Security Measures for as long as it is retained.
2.6 Language. Databeskyttelsesloven is authentic in Danish. The descriptions in this Section are provided in English for convenience; in the event of a discrepancy, the Danish text of the statute governs.
3. NIS2 Directive
3.1 Regulated Entity Status. SDU eScience operates services falling within the scope of NIS2. SDU eScience Center is a department of an institute at a faculty of the University of Southern Denmark (Syddansk Universitet) and is not a separate legal entity. The registration under the Danish NIS2 Implementation is accordingly held by the University of Southern Denmark, which is registered as an essential entity, and that registration covers the Services. SDU eScience provides the Services subject to the supervisory regime that the Danish NIS2 Implementation applies to essential entities. Where this Section 3 states an obligation of SDU eScience, that obligation is one of the University of Southern Denmark discharged in respect of the Services.
3.2 Cybersecurity Risk-Management Measures. SDU eScience maintains cybersecurity risk-management measures appropriate to the risks posed to the security of its network and information systems, addressing at least the matters listed in Article 21(2) of NIS2, namely:
(a) policies on risk analysis and information system security;
(b) incident handling;
(c) business continuity, including backup management and disaster recovery, and crisis management;
(d) supply chain security, including security-related aspects of the relationships between SDU eScience and its direct suppliers and service providers;
(e) security in the acquisition, development and maintenance of network and information systems, including vulnerability handling and disclosure;
(f) policies and procedures to assess the effectiveness of the cybersecurity risk-management measures;
(g) basic cyber hygiene practices and cybersecurity training;
(h) policies and procedures on the use of cryptography and, where appropriate, encryption;
(i) human resources security, access control policies and asset management; and
(j) the use of multi-factor authentication or continuous authentication solutions, and of secured communications and emergency communication systems.
SDU eScience maintains the Compliance Certifications described in Section 7.4 (Compliance Certifications) of the Data Processing Terms, which include certification to ISO/IEC 27001 for the Audited Services, as listed at https://legal.cloud.sdu.dk/security/compliance/services-in-scope. The measures described in this Section are implemented, reviewed and audited within the information security management system that certification covers.
These measures operate alongside, and do not reduce, the Security Measures described in Appendix 2 to the Data Processing Terms.
3.3 Governance. The management body of the University of Southern Denmark is the management body responsible for the purposes of Article 20 of NIS2. It approves the cybersecurity risk-management measures described in Section 3.2 as they apply to the Services, oversees their implementation, and undertakes the training contemplated by that Article.
3.4 Incident Reporting to Authorities. Where a significant incident within the meaning of Article 23 of NIS2 occurs, SDU eScience will submit to the competent authority and to the CSIRT designated under the Danish NIS2 Implementation:
(a) an early warning without undue delay and in any event within 24 hours of becoming aware of the significant incident;
(b) an incident notification without undue delay and in any event within 72 hours of becoming aware of the significant incident, updating the early warning and providing an initial assessment, including severity, impact and any indicators of compromise;
(c) an intermediate report on the request of the CSIRT or competent authority; and
(d) a final report not later than one month after the submission of the incident notification under subsection (b).
Reporting under this Section concerns the network and information systems on which the Services are provided, and is a legal obligation of the University of Southern Denmark as the registered essential entity, discharged in respect of the Services by SDU eScience. SDU eScience will include Customer Personal Data in such a report only to the extent the Danish NIS2 Implementation requires, and will otherwise report in aggregated or anonymized form.
3.5 Notification to Customer. Where a significant incident reported under Section 3.4 affects the Services provided to Customer, SDU eScience will notify Customer without undue delay, using the Notification Email Address, and will describe the nature of the incident, the measures taken or planned, and any measures SDU eScience recommends Customer take. Where the incident is capable of adversely affecting the provision of the Services, that notification will be given even if the incident does not involve Customer Personal Data.
3.6 Relationship to Data Incident Notification. A single event may be both a Data Incident under the Data Processing Terms and a significant incident under NIS2. Notification under one does not discharge the obligation to notify under the other, and the applicable deadlines run independently. Notification under this Section is not an acknowledgement by SDU eScience of fault or liability.
3.7 Customer’s Own Obligations. Where Customer is itself an essential or important entity under NIS2 or equivalent national law, Customer remains responsible for its own risk-management measures and for its own reporting obligations. On Customer’s reasonable request, and taking into account the information available to it, SDU eScience will provide information about the Services that Customer reasonably requires in order to meet its supply chain security obligations under Article 21(2)(d) of NIS2. SDU eScience may charge a reasonable fee for assistance that goes beyond the provision of its standard Security Documentation.
3.8 Vulnerability Disclosure. SDU eScience maintains a coordinated vulnerability disclosure process. Customer will report any vulnerability it identifies in the Services to SDU eScience through that process rather than disclosing it publicly, and will not use a vulnerability it identifies to access data other than its own.
4. EU AI Act
4.1 SDU eScience’s Current Role. SDU eScience provides technical infrastructure: computing, storage and networking capacity on which Customer runs its own workloads. Except as provided in Section 4.8 for a Service identified as an AI Service in the Service Specific Terms, SDU eScience does not place an AI system or a general-purpose AI model on the market or put one into service, and is therefore not a provider or deployer of AI systems within the meaning of the AI Act in respect of Customer’s use of the Services.
SDU eScience does not monitor, review, classify or assess Customer’s AI systems, Customer’s use of the Services, or Customer’s compliance with the AI Act, and has no obligation to do so. SDU eScience does not determine the purpose for which an AI system is developed, trained or used, exercises no control over its design, training data or outputs, and is not responsible for those outputs or for any decision taken on the basis of them. This paragraph is subject to Section 4.8.6, which describes the only records SDU eScience keeps of Customer’s use of an AI Service.
4.2 Customer’s Role. Where Customer develops, trains, deploys or makes available an AI system or a general-purpose AI model using the Services, Customer is the provider or deployer of that AI system or model, and is responsible for compliance with the AI Act in that capacity. This includes determining the risk classification of the AI system, and meeting any obligations that classification attracts.
4.3 Prohibited Practices. Customer will not use the Services to develop, train, or operate an AI system for any practice prohibited by Article 5 of the AI Act. This restriction is additionally reflected in the Acceptable Use Policy. It is a contractual restriction on Customer and does not oblige SDU eScience to monitor for, detect or prevent such use. Where an AI Service is provided, the records described in Section 4.8.6 may bring a breach to SDU eScience’s attention, but keeping those records does not create an obligation to monitor for, detect or prevent a breach, and SDU eScience does not warrant that they will reveal one. Where SDU eScience becomes aware of a breach, its rights are those set out in Section 4 (Suspension) of the Terms of Service.
4.4 General-Purpose AI Models and Systemic Risk. A general-purpose AI model is presumed under the AI Act to present systemic risk where the cumulative amount of computation used for its training exceeds the threshold set out in Article 51 of the AI Act. Customer will notify SDU eScience in advance where Customer intends to use the Services for a training run that may meet or exceed that threshold, so that SDU eScience can plan capacity. Notification is for operational purposes only. It does not make SDU eScience the provider of the model, does not oblige SDU eScience to determine whether the threshold has been met, and does not relieve Customer of its own obligation to notify the European Commission under the AI Act.
4.5 AI Literacy. Each party is responsible under Article 4 of the AI Act for ensuring a sufficient level of AI literacy among its own staff and other persons operating AI systems on its behalf.
4.6 Terminology. The term “High Risk Activities” as defined in the Terms of Service has a meaning unrelated to the term “high-risk AI system” as used in the AI Act. A restriction on High Risk Activities in the Terms of Service is not a restriction on high-risk AI systems, and neither term should be read by reference to the other.
4.7 Records and Cooperation. Where a competent authority under the AI Act requests information about Customer’s use of the Services, SDU eScience will disclose that information only to the extent required by applicable Legal Process, and will disclose no more than that Legal Process requires; provided that SDU eScience uses commercially reasonable efforts to: (i) promptly notify Customer of such disclosure before disclosing; and (ii) comply with Customer’s reasonable requests regarding its efforts to oppose the disclosure. Notwithstanding the foregoing, subsections (i) and (ii) above will not apply if SDU eScience determines that complying with (i) and (ii) could: (a) result in a violation of Legal Process; (b) obstruct a governmental investigation; and/or (c) lead to death or serious physical harm to an individual. This Section applies the same approach as Section 8.2 (Required Disclosure) of the Terms of Service. Where the information requested includes Customer Personal Data, Section 5.2.5 (Processing Required by Law) of the Data Processing Terms also applies.
Customer remains responsible for responding to any request addressed to it in its capacity as provider or deployer, and SDU eScience will refer such requests to Customer where it is permitted to do so.
4.8 AI Services Provided by SDU eScience. This Section applies where SDU eScience makes available a Service that is an AI system, or that makes a general-purpose AI model available for inference, and that is identified as such in the Service Specific Terms (an “AI Service”). Sections 4.1 and 4.2 continue to govern every other Service. Terms defined in the Service Specific Terms for an AI Service, including “Model”, “Prompt”, “Model Output”, “Model Safeguards” and “Service Safeguards”, have the same meaning in this Section.
4.8.1 SDU eScience’s Role. In respect of an AI Service, SDU eScience is the provider of the AI system made available through that Service within the meaning of Article 3(3) of the AI Act: SDU eScience develops that system, incorporates into it Models developed by third parties, and puts it into service under its own name. SDU eScience is not the provider of any Model made available through an AI Service unless it places that Model on the market or puts it into service under its own name or trademark, or substantially modifies it, within the meaning of Article 25 of the AI Act.
4.8.2 No Substantial Modification. SDU eScience makes each Model available for inference with the Model Safeguards implemented by the provider of that Model intact. SDU eScience will not remove, disable, relax or bypass Model Safeguards, and will not make available through an AI Service a Model from which Model Safeguards have been removed, disabled or relaxed. Service Safeguards operate only to restrict Model Output and not to extend it. Making a Model available on those terms is not a substantial modification of the Model, and does not make SDU eScience the provider of the Model or of a general-purpose AI model. Where SDU eScience is unable to satisfy itself that a Model’s Model Safeguards are intact, it will not make that Model available through an AI Service.
4.8.3 Customer’s Role. Where Customer integrates an AI Service, or Model Output, into an AI system that Customer places on the market or puts into service, Customer is the provider of that AI system. Where Customer uses an AI Service under its own authority, Customer is a deployer of that AI system. Section 4.2 applies to Customer in either capacity. Where Customer places an AI Service on the market under its own name or trademark, or substantially modifies it, Article 25 of the AI Act applies to Customer.
4.8.4 Prohibited Practices. SDU eScience will not put an AI Service into service for a practice prohibited by Article 5 of the AI Act. Model Safeguards and Service Safeguards are not a means by which SDU eScience assesses, or is able to assess, whether Customer’s use of an AI Service is such a practice, and their operation is not a representation that it is not. Section 4.3 applies to Customer’s use of an AI Service.
4.8.5 Transparency. Where Article 50(2) of the AI Act requires the output of a generative AI system to be marked in a machine-readable format as artificially generated or manipulated, SDU eScience will meet that obligation for an AI Service at the level of the Service: the response returned by the Service identifies, in machine-readable form, that the output was generated by the identified Model. Customer will not remove, alter or suppress that marking. The obligations under Articles 50(1), 50(3) and 50(4) of the AI Act, including informing a natural person that they are interacting with an AI system and disclosing deep fakes and artificially generated text published with the purpose of informing the public on matters of public interest, fall on Customer as deployer where Customer makes an AI Service or Model Output available to others. SDU eScience does not discharge those obligations on Customer’s behalf.
4.8.6 Records of Safeguard Refusals. SDU eScience records events in which a Model Safeguard or a Service Safeguard refuses or withholds Model Output, as described in the Service Specific Terms for the AI Service. The record includes the Prompt that gave rise to the refusal, is kept only in order to detect and investigate breaches of the AUP or of Section 3.3 (Restrictions) of the Terms of Service and to exercise SDU eScience’s rights under Section 4 (Suspension) of the Terms of Service, and is retained for the period stated in those Service Specific Terms. Section 5.2.4(c) of the Data Processing Terms and Appendix 1 to those Terms apply to that processing. SDU eScience is under no obligation to review a record it has created.
4.8.7 No Assurance of Compliance. The operation of Model Safeguards or Service Safeguards on an AI Service does not: (a) discharge any obligation of Customer as a provider or deployer under the AI Act; (b) constitute human oversight within the meaning of Article 14 of the AI Act, or any part of the measures a deployer must take under Article 26 of the AI Act; (c) make an AI Service suitable for use as, or as a component of, a high-risk AI system within the meaning of Article 6 of the AI Act, or for High Risk Activities as defined in the Terms of Service; or (d) make Model Output lawful, accurate, or fit for any purpose.
4.8.8 Model Licence Terms. A Model made available through an AI Service may be subject to licence terms or an acceptable use policy of the provider of that Model. SDU eScience will make those terms available to Customer, and Customer will comply with them in its use of the AI Service. Where those terms and the AUP differ, the more restrictive applies.
4.8.9 Information for Customer. On Customer’s reasonable request, SDU eScience will provide the information about an AI Service that Customer reasonably requires in order to meet its own obligations as a provider or deployer under the AI Act, to the extent that information is available to SDU eScience and its disclosure is permitted by the licence terms for the Model. SDU eScience does not carry out a conformity assessment, a fundamental rights impact assessment under Article 27 of the AI Act, or a classification of Customer’s AI system, and this Section does not oblige it to do so.
4.9 Reserved.