Designing trust with software agents A case study INTRODUCTION Secure interchange of information can nowadays be supported by well known and well understood software components. Familiar solutions include digital signatures, encryption and hashing. When two parties, unknown to each other, want to safe- guard privacy and trust in their communication and transactions, we enter into a more abstract level of security. A trusted third party, if there is one, can support the communication in this case. It mimics real world situations. An even bigger challenge is to build trust and privacy into a software system that lacks trusted third parties. In this case, software agents can bridge the gap between existing security technology and the user requirements. The chal- lenge then is to incorporate two levels of trust: Trust questions which exist in the real world have to be modelled in the software agents, and the sys- tem in itself must show trustworthiness to the involved parties. In this paper we intend to reflect upon these trust questions. After a presentation of software agent technology as a further development of computer Info, Comm & Ethics in Society (2006) 1: 37-48 © 2006 Troubador Publishing Ltd. Stijn Bernaer, Martin Meganck and Greet Vanden Berghe KaHo St.-Lieven, Department of Industrial Engineering, Gebr. Desmetstraat 1, 9000 Gent, Belgium Email: {stijn.bernaer,martin.meganck,greet.vandenberghe}@kahosl.be Patrick De Causmaecker Katholieke Universiteit Leuven, Campus Kortrijk, Computer Science and Information Technology, E. Sabbelaan 53, 8500 Kortrijk, Belgium Email: Patrick.DeCausmaecker@kuleuven-kortrijk.be In this paper, we will address privacy and trust issues that arise in more advanced software systems. Though a lot of information is currently available in electronic form, not all of it should widely be accessible to everybody. The involved parties need full control on how their data are used and who has access. If the system consists of autonomous software agents, this problem requires extra attention and new working principles. We illustrate this in the case of a communi- cation platform for multimodal transport. The major aim of the communication platform is to enhance exchanging infor- mation and to ultimately improve organisation/collaboration within the transport sector. A better informed view of the transport sector will facilitate better considered decisions for users of the communication platform. The software sys- tem merits credibility by accurately modelling all the relevant real world interactions of potential users of the system. We opted for a connectivity solution in which software agents act as representatives of the parties involved. All agents can be equipped with human-like skills and qualities such as intelligence, autonomy, and the ability to cooperate, coor- dinate and negotiate. We demonstrate how cooperation between parties can be achieved while respecting their sensi- tivity concerning information. Keywords: privacy, trust, software agents VOL 4 NO 1 JANUARY 2006 37