VIKING
Trustworthy artificial intelligence for potential police applications
In complex police investigations – including cases of organized crime or terrorism – large volumes of digital data are regularly generated, in particular images, videos, audio recordings, text and voice messages. Given the limited human resources available, the timely manual analysis of such data poses extreme challenges for investigative teams. AI solutions promise to remedy this situation by enabling presorting and filtering of the data according to their potential relevance, recognizing cross matches or pointing out hidden connections in the data thicket.
A fundamental contribution of VIKING is the development of a catalogue for compliance with acceptable ethical and high legal requirements for AI procedures; it operationalizes the underlying principles of trustworthy AI in the light of the new EU law and further legal and ethical requirements. This guideline will differentiate between the requirements for providers and users in a stakeholder-friendly manner.
With the interlocking of different activities in the VIKING project, the foundations for trustworthy AI will be created. Users will accompany the work and evaluate the demonstrators and requirement catalogues with regard to functionality and practicability for everyday police use, which ensures that the research work is closely linked to actual relevance for the users. If successful, the results from VIKING can advance to best practice for police use of AI procedures in the future, strengthen law and security in Europe and significantly shape the rapidly growing national and international markets in this segment with technology made in Germany.
A practice-relevant contribution of VIKING is the knowledge transfer of the project results into standardization activities, which aim at the developer-independent measurability and verifiability of AI solutions taking into account acceptable ethical and high legal requirements and form the basis of future certification procedures. The standardization documents developed during the funding period, such as DIN SPECs, are intended to serve as a basis for European and international standardization activities in the long term and form a possible groundwork for Standards in the field of civil security and artificial intelligence.
Project duration
January 2022 - January 2025
Project coordinator
IDEMIA Identity & Security Germany AG
Project partners
- Universität Konstanz
- Zentrale Stelle für Informationstechnik im Sicherheitsbereich
- Fraunhofer Institut für digitale Medientechnologie - Institutsteil Hör-, Sprach- und Audiotechnologie
- Universität Tübingen
- Hochschule für Wirtschaft und Recht Berlin, Forschungsinstitut für Öffentliche und Private Sicherheit (FÖPS Berlin)
- Landeskriminalamt Nordrhein-Westfalen
- Universität der Bundeswehr München – Forschungsinstitut CODE
- Polizeipräsidium München
- DIN Deutsches Institut für Normung e. V.
Funding
The VIKING project is funded by the Federal Ministry of Education and Research within the framework of the call "Artificial Intelligence in Civil Security Research II under the funding code: 13N16240.
For further information please see:
www.sicherheitsforschungsprogramm.de
