Criminal Liability and 'Smart' Environments

Smart applications are said to be adaptive to differing circumstances. They ‘learn’ about the best way to achieve the goals for which they have been programmed. Smart environments are said to be proactive with regard to the preferences of their users and the risks they run. Ambient Intelligence, ubiquitous, pervasive or autonomic computing all concern visions of a new socio-technical infrastructure that manages to always stay one step ahead of its human users. These environments are meant to take a whole range of decisions, having real consequences for their ‘users’. The ‘intelligence’ that is said to enable ‘smart’ decision-making thrives on the connectivity between different devices and cannot always be traced to one particular device or to one particular natural or legal person. It seems that we have a new kind of collective ‘agency’ here that may be entirely nonhuman or a complex hybrid of human and nonhuman entanglements. If these nonhuman or hybrid ‘agents’ cause harm to others a series of questions come to mind: to what extent can such agency be identified as ‘causing’ harm; what is the meaning of ‘intent’, ‘intention’ and ‘mens rea’ in the case of nonhuman agency; do we need to revise our conceptions of agency, causality, intent, intention and mens rea to accommodate the fact that nonhuman or hybrid ‘agents’ do not fit our traditional conceptual framework, or, alternatively, should we reject applicability of the criminal law in cases that fall outside the scope of human intention as we presently understand it. In the case we opt for non-applicability of the criminal law the question is raised of how will we deal with harm caused for which we cannot attribute criminal liability, due to the fact that the technological infrastructure seems to have ‘caused’ the harm. This, in its turn, requires a discussion of whether the ‘acts’ of the technological architecture will have a status similar to ‘acts of Nature’ or ‘acts of God’.