Conclusions, August 2003
analysis of legal and technical implications of the use of software agents
Goal of the research
The goal of this interdisciplinary project is to explore the legal implications of agent technology, specifically focussing on implications for software development.
The concrete aim of this project was to provide a 'cookbook' for (1) designers of agent systems on the way to design (and code) intelligent agents in a legally valid way and (2) legal experts, policy makers as well as other relevant parties indicating where new laws have to be formulated or clarification is needed with respect to existing law. This cookbook has now been published as a technical report: see http://www.iids.org/alias/
Process and Results
This project required considerable interaction between legal experts, computer scientists and AI experts. To acquire a common understanding of the problem area and its breadth a number of scenarios were studied: the goal being to understand the issues involved. These scenarios include grocery shopping, chemical commodities marketplace, local government automation and hospital automation. Relevant legal and technical implications have been analysed, e.g., resulting in the realisation that a conceptual distinction between open and closed systems provides a basis for further analysis.
Intermediary concepts were developed to overcome language and conceptual barriers between legal experts and computer scientists, forming the basis for a structured analysis of legal aspects of software agents. The areas explored included
- agent autonomy with e.g. respect to contract negotiation, eg Are agents legal entitites? Who's liable for possible damage?
- identifiability and traceability, eg Can an agent remain anonymous? Pseudo-anonymous?
- integrity and originality, eg: May sysadmins kill agents on their own machines? With which risk?
- trust, eg: What are the obligations of an agent with respect to confidentiality?
Conclusion
The results of this project show that this area is clearly very broad, the legal aspects are [numerous and are not yet fully clear, nor are they] fully understood, and that interaction between different disciplines is necessary will they ever be understood. This interaction will help legal experts to adapt present law so new technologies can be lawfully applied. It will also help software designers to comply with the legal requirements.
Identity management is one of the areas in which further research is required: technological and legal aspects need to be considered in this process. The researcher groups involved in this project are continuing this research to acquire a better understanding of the options and their implications.