Digital ethics is a current and much-described topic. National and international players from politics and industry have developed guidelines, codes or charters that define digital ethics and declare the intention to act according to certain values and principles.
At the same time, more and more projects are emerging that go one step further and try to operationalize these principles. One approach to this is Digital Trust Labels, which certify websites, software applications and online services. Similar to the food industry (e.g. organic label), the aim of such labels is to identify products for end users and/or business partners in such a way that informed purchasing or usage decisions can be made quickly.
In the following, we present a selection of interesting digital trust labels and certificates that have been published so far.
Open Ethics Initiative | Open Ethics Label
Inspired by the food industry, the Open Ethics Initiative has launched the Open Ethics Label. The label aims to strengthen users' trust in artificial intelligence through transparency. Software developers of AI systems can apply for the label at the global, inclusive initiative and will be assessed on three key factors: Training Data (How does the system use the data?), Algorithms (How is the data processed by the system?) and Decision Space (What decisions are made by the system?).
Data Nutrition Project | The Dataset Nutrition Label
Based on the nutrition labelling of food in the North American market, the Dataset Nutrition Label was developed by the US Data Nutrition Project. The aim is to create greater transparency as to how a particular dataset is structured and what risks may be associated with its use. These risks can be, for example, bias or prejudice in data sets, or a lack of ethical review of the data. A data scientist looking for a data set can now refer to this label and identify possible risks at an early stage. With its work, the project team wants to promote progress through technology and prevent existing systemic injustice from being reflected in it.
AI Ethics Impact Group, Bertelsmann Stiftung | From Principles to Practice – An interdisciplinary framework to operationalise AI ethics
In 2020, AI Ethics Group, an interdisciplinary consortium led by VDE e.V. and the Bertelsmann Foundation, published its operationalization framework for AI ethics. This contains, among other things, an AI Ethics Label, which visually and structurally represents the ethical rating of an AI system, based on the representation of energy efficiency classes. Ratings from A to G are assigned in six categories, including transparency, accountability, privacy, justice, reliability and environmental friendliness.
From Principles to Practice - An interdisciplinary framework to operationalize AI ethics (PDF)
EU Next Generation Internet Initiative | Trustmark for the Internet
The Trustmark for the Internet is a label established in 2020 by the EU's NEXT GENERATION INTERNET initiative. It is aimed at websites, online services and apps and checks them against certain criteria. These criteria include cyber security, privacy and data protection, transparency, bias and prejudice, accountability and sustainability. To present the complex results of the analysis in a quick and easy-to-understand way, a traffic light display with the possibility of calling up further information will be used. The aim of the initiative is to highlight providers that have been checked with the label and to make them more visible to users.
Trustmark for the Internet (PDF)
German Institute for Standardization (DIN) | Standards Roadmap AI
Due to the lack of reliable quality criteria and test procedures for AI systems, the German Institute for Standardization has launched a standardization roadmap for AI in 2020. With the help of this roadmap, standardized test procedures and benchmarks for AI systems are to be developed. Passing these tests and procedures by AI systems will allow them to certify a certain standard and attest to conformity. Since there is no comparable solution in the world yet, the German Institute for Standardization has set itself the goal of developing one.
Standardization roadmap AI (PDF)
Danish government & The Danish Industry Foundation | D-seal: Seal for Data Ethics
As a motivation for responsible data handling, the Danish government together with The Danish Industry Foundation launched the D-seal in 2019. The label is intended to give participating companies a competitive advantage by certifying the ethical and trustworthy handling of data. A total of 8 different criteria are examined, such as whether top management takes responsibility for IT security and data use and whether users are informed about what data they need and what it is used for. Furthermore, the algorithms used as well as the distribution of data and its use to and by subcontractors will also be examined. In addition to the companies, the users should also benefit from the label as an orientation aid.
Fraunhofer IAIS | Trustworthy Use of Artificial Intelligence
AI applications bring with them a disruptive potential, which means that compliance with philosophical, ethical and legal frameworks must be ensured. For this purpose, Fraunhofer IAIS has developed an AI certification in 2019 which examines six basic values and their implementation and compliance in AI applications under the umbrella of Ethics & Law. These include, for example, autonomy & control, fairness, reliability and transparency. The resulting test catalogue enables accredited examiners to assess AI applications in an expert and neutral manner.
Trustworthy use of artificial intelligence (PDF)
Guideline for the design of trustworthy artificial intelligence - AI test catalogue (PDF)
Agencie-Lucie | Le label Numérique Responsible
The NR label, launched by the French Agencie-Lucie, deals with the digital ecological footprint of organizations. The digital transformation is accompanied by major environmental challenges that are still too seldom the focus of social discussion and too little addressed. Due to the boom in energy consumption, the increase in greenhouse gas emissions, soil and air pollution and the use of non-renewable resources, this footprint is constantly growing. The NR label recognizes organizations that use responsible digital technology and aim to reduce the environmental, economic and social footprint of information and communication technologies. The NR label is also supported by French public institutions, universities and companies from the private sector.
Le label Numérique Responsable
Collective effort: Consumer Reports, Disconnect, Ranking Digital Rights, The Cyber Independent Testing Lab | The Digital Standard
The Digital Standard was created in 2018 by a consortium of several non-profit organizations. The task of the framework is to measure consumer values such as privacy, security and property and to define best practices in dealing with these. For the security aspect, emphasis is placed on ensuring that electronic devices and software are built securely and that the risk of malware, for example, is reduced. The goal for the consumer value of privacy is that this is preserved for the customer and that stored customer data is secured and stored in a monitored manner. The product ownership regime should allow the consumer to resell, modify or repackage the product after purchase. The framework is supplemented by governance rules that state that companies should behave ethically and protect freedom of expression with their products, among other things.
