Data Privacy Notice

Data Processor

The responsible party for data processing within the meaning of Art. 4 No. 7 GDPR and other national data protection laws of the member states as well as other data protection regulations is the:

Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen
Burckhardtweg 4
37077 Göttingen
Göttingen, Germany
Tel: +49 (0) 551 39-30001
E-mail: support@gwdg.de
Website: www.gwdg.de

Represented by the managing director. The controller is the natural or legal person who alone or jointly with others determines the purposes and means of the processing of personal data.

Contact person / Data protection officer

Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen
Datenschutzbeauftragter
Burckhardtweg 4
37077 Göttingen
Göttingen, Germany
Phone: +49 (0) 551 39-30001
E-mail: support@gwdg.de

General information on data processing

Overview of the service

The ChatAI service consists of several components, in particular a web frontend and a large language model in the backend. The front end offers users a web interface to enter user requests directly via a browser. Furthermore, the desired model can be selected and certain settings can be made. The frontend forwards all requests to the selected model backend. For data protection reasons, a distinction is made here between models hosted locally by the GWDG itself and external models from other providers, whereby the external models are labelled as such. The backend is hosted via the GWDG platform SAIA, it receives all requests and forwards them to the corresponding model. In the case of external models, the requests, or more precisely, the history stored in the browser, including the intermediate texts of the models as made by the user, are forwarded to the external provider. In the case of self-hosted models, the requests are only processed on the GWDG systems.

Scope of the processing of personal data

We only process our users’ personal data to the extent necessary to provide a functional website and our content and services. The processing of personal data of our users takes place regularly only with the consent of the user (Art. 6 para. 1 lit. a GDPR). An exception applies in cases where prior consent cannot be obtained for factual reasons and the processing of the data is permitted by law.

Insofar we obtain consent from the data subject for processing operations involving personal data, Article 6 (1) lit. (a) of the EU General Data Protection Regulation (GDPR) is the legal basis for personal data processing.

When processing personal data that is necessary for the performance of a contract to which the data subject is party, Article 6 (1) lit. (b) GDPR is the legal basis. This also applies to processing operations that are necessary for the performance of pre-contractual measures.

Insofar as the processing of personal data is necessary for compliance with a legal obligation to which our company is subject, Article 6 (1) lit. (c) GDPR is the legal basis.

Where the processing of personal data is necessary in order to protect the vital interests of the data subject or another natural person, the legal basis is Article 6 (1) lit. (d) GDPR.

If the processing is necessary to protect a legitimate interest of our company or a third party and the interests, fundamental rights and freedoms of the data subject do not outweigh the first-mentioned interest, Art. 6 (1) lit. f DSGVO is the legal basis for the processing.

Use of the Chat-AI website (frontend)

Description and scope of data processing

Each time https://chat-ai.academiccloud.de/ is accessed, the system automatically collects data and information from the computer system of the accessing computer. The following data is collected in each case:

  • Date of access
  • Name of the operating system installed on the accessing device
  • Name of the browser used
  • Source system via which the access was made
  • The IP address of the accessing device

The data is also stored in the log files of our system. This data is not stored together with other personal data of the user. All Chat-AI data displayed in the browser is only stored on the client side in the user’s browser and is only transmitted to the server for the necessary processing when the user requests it, i.e. while the data is being processed by the backend models. After the end of a session in the browser, no more user input is available.

General use of models

Description and scope of data processing

For billing purposes, the following data is stored and logged on the GWDG server for each request:

  • Date of the request
  • user ID
  • Length of the request and response

This data is also stored in the log files of our system. This data is not stored together with other personal data of the user. Depending on whether locally hosted models or external models are used, slightly different data protection provisions apply. No liability can be accepted for the automatically generated answers. Answers may be completely incorrect, contain incorrect partial information or may have unlawful content.

Duration of storage

The billing data is stored for one year.

Use of self-hosted models

Description and scope of data processing

In order to use the models hosted by the GWDG, the user’s input/requests are processed on the GWDG’s systems. Protecting the privacy of user requests is of fundamental importance to us. For this reason, our service in combination with the self-hosted models does not store the contents of the requests (chat history), nor are requests or responses stored on a permanent memory at any time.

Duration of storage

The entries are only stored on the GWDG server during processing by the Large Language Models themselves, i.e. while the data is being processed on their own systems.

Use of external models from OpenAI

Description and scope of data processing

In order to use the OpenAI models, we send the respective request (user input) from our server to the Microsoft servers (external service provider). The following data is forwarded to fulfil the service

  • User request

Information about the users themselves is not forwarded by GWDG. However, the user’s enquiry is forwarded unfiltered, i.e. personal information contained in the enquiry itself is forwarded to the external service provider. The GWDG service is based on the Data Processing Addendum ( https://learn.microsoft.com/en-us/legal/cognitive-services/openai/data-privacy, https://www.microsoft.com/licensing/docs/view/Microsoft-Products-and-Services-Data-Protection-Addendum-DPA). This means that there is an adequacy decision in accordance with the European General Data Protection Regulation, but data transmission to third parties cannot be ruled out by GWDG.

The requests are anonymised by the GWDG servers and are send to the external service provider where they are only logged for up to 30 days in accordance with the Microsoft Data Processing Addendum in the event of an attempt at misuse, e.g. to create hate or sexualised content. This happens automatically if the backend detects an attempt at misuse. It cannot be ruled out that legitimate requests may be incorrectly categorised and logged as attempted abuse.

Possibility of objection and removal

The recording of the user’s input and the processing by Microsoft is mandatory for the provision of the external models. Consequently, there is no possibility for the user to object.

Use of Research Partner Models

Description and Scope of Data Processing

GWDG has research partners who host models externally on their compute resources. For this purpose, GWDG forwards the corresponding user request to the research partners. Information about the users themselves is not forwarded. However, the user requests are forwarded unfiltered, i.e. personal data contained in the request itself is forwarded to the research partners. The GWDG has a research contract with the relevant research partners. Corresponding models that are hosted by research partners are marked within the ChatAI web interface by a following “(Research Partner)”. The use of these models is at your own risk.

Use of models hosted at LUIS

Leibniz University IT Services (LUIS) operates an inference cluster as part of the KISSKI project, which host some large language models. These models are labelled with “LUIS” in the menu of the ChatAI service.

Description and scope of data processing

In order to use the models hosted by LUIS, the user’s input/requests are processed on the LUIS systems. Protecting the privacy of our users data is of fundamental importance to us. For this reason, our service, in combination with the models hosted at LUIS does not store the chat history nor are requests or answers stored on a permanent storage at any time.

Duration of storage

The input is only stored on the servers hosted by LUIS during processing of the language models, i.e., while a response to the request is being generated.

Use of the server-side RAG system

Description and scope of data processing

Generally, a basic distinction has to be made between the role of the developer and the user. The developer provides context data that is used to build an index on the server side. This index is persistent and can be used across multiple sessions and by multiple users. This index is used to give users access to a large language model that can use specific knowledge from the provided context data to answer the users’ individual queries. For this, the developer uploads the data via the RAGManager, where it is stored and indexed in different data sets, called Arcanas. Access to an Arcana is secured with a password. An Arcana can be shared with any number of people. Each user must know the name of the Arcana, or rather its ID, and the corresponding password it is secured with. It is important to highlight that anyone who has access to the password can access the Arcana and therefore the knowledge it contains. An Arcana can only be used in combination with an open source model hosted by the GWDG. This feature is not available in the external models, which are explicitly identified as such in the name.

Description and scope of data processing

The context data provided by the developers is indexed into an Arcana on the server side and secured with a password. The Arcana is then exported to the context of the open source models in Chat AI if an ID and the correct password have been provided by the user.

Duration of storage

The context data provided by the developers is stored permanently until it is explicitly deleted by the developers. The requests and responses for the users are still only stored locally on the users’ client systems, as described in the section “Use of self-hosted models”. The users requests is only available on the GWDG servers while it is being processed.

Rights of data subjects

You have various rights with regard to the processing of your personal data. We list them in the following, but there are also references to the articles (GDPR) and/or paragraphs (BDSG (2018)) which provide even more detailed information.

Right of access by the data subject (Article 15 GDPR; § 34 BDSG)

You may request confirmation from the controller whether we process personal data related to you. This includes the right to obtain access to information as to whether the personal data concerning you is transferred to a third country or to an international organisation.

Right to rectification (Article 16 GDPR)

You have a right of rectification and / or completion vis-à-vis the controller if the personal data processed related to you is inaccurate or incomplete. The controller must perform rectification immediately.

Right to erasure / “Right to be forgotten” / Right to restriction of processing (Article 17/18 GDPR; § 35 BDSG)

You have the right to request the immediately erase of your personal data from the controller. As an alternative, you may request to restrict the processing from the controller, whereby restrictions are referred to in the GDPR/BDSG under the articles and/or sections mentioned.

Notification obligation regarding rectification or erasure of personal data or restriction of processing (“Right to be informed”) (Article 19 GDPR)

If you have asserted the right to rectification, erasure or restriction of processing vis-à-vis the controller, the controller is obligated to communicate such rectification or erasure of the data or restriction of processing to all recipients to whom the personal data concerning you has been disclosed, unless this proves impossible or involves disproportionate effort. You have the right vis-à-vis the controller to be informed about these recipients.

Right to data portability (Article 20 GDPR)

You have the right to receive the personal data concerning you, which you have provided to the controller, in a structured, commonly used and machine-readable format. In addition to the scenarios presented in and provisions of the GDPR, it must be noted that portability of mass data / user data is limited to technical readability. The right to data portability does not include that the data created by the user in a proprietary format is converted by the controller into a commonly used, i.e. standardised format.

Right of objection (Article 21 GDPR; § 36 BDSG)

You have the right to object to the processing if this is based only on the controller weighing any interests (see Article 6 (1) lit. (f) GDPR). Right to withdraw consents in terms of data protection laws (Article 7 (3) GDPR) You have the right to withdraw your consent under data protection laws at any time. The withdrawal of consent does not affect the lawfulness of processing based on such consent before its withdrawal.

Right to complain to a supervisory authority (Article 77 GDPR)

Without prejudice to any other administrative or judicial remedy, you have the right to lodge a complaint with a supervisory authority, in particular in the Member State of your habitual residence, place of work or place of the alleged infringement if you consider that the processing of personal data relating to you infringes the GDPR.