MVV347K AI, Law, and Governance
Lecture 5 – AI and liability
This session will address the use of data in AI systems and application and the legal
implications in terms of legal status, responsibility, and legal liability.
Question: Should we introduce legal personality for AI in order to grant them rights and
responsibilities?
AI systems are currently not assigned any specific legal status in law.
We need to ask, what would be the reason to change that?
What are the technical, legal, and ethical justifications for creating a legal personality for
autonomous intelligent agents?
Which legal theory could support his and what would be the practical implementations?
The legal status would entail a set of rights and responsibilities that would reflect the nature
of the entity – do we truly understand the nature of autonomous intelligent agents now and
in the future?
This links back to the discussion on transparency, explainability and interpretability
The importance of metaphors – how do we see AI and robots? As killers, sophisticated tools,
or electronic slaves?
Recommended reading: Ugo Pagallo, ‘Killers, fridges, and slaves: A legal journey in robotics’
Chris Reed, Elizabeth Kennedy, Sara Nogueira Silva, ‘Responsibility, Autonomy and
Accountability: legal liability for machine learning’
Task: Consider the distinction between responsibility, accountability, and liability in legal
context and how it relates to autonomous intelligent agents. Can we attach either of
these concepts to them? Is there a useful analogy, such as children, animals, or slaves in
Roman law?
Criminal and civil (private) law perspectives
As the law currently stand, non-human agents do not exhibit the necessary elements
required by criminal law, such as consciousness and intention
Private law – contractual and ex-contractual liability – autonomous intelligent agents can act
on behalf of their masters (principals) and create rights and responsibilities in their benefit
(or detriment)
Further reading: Ugo Pagallo, ‘The Laws of Robots: Crimes, Contracts, and Torts’
We can consider the relationship of liability of autonomous intelligent agents, legal and
practical implications in the following areas:
• Big data and data protection
• Predictive analytics and privacy
• Access to and control of data
• Market competition
Task: Consider the role of contract law and insurance to manage risk for developers and
other participants.
Can we draw parallels with other scenarios from Cyberspace where overreliance on
contractual or technological mechanisms led to a significant social, legal and economic
negative externality?
Digital peculium
Concept based on Roman Law
Digital Peculium is a special set of rules that would define the parameters of liability for
autonomous agents in the context of commercial transactions
Recommended reading: A Katz, M MacDonald, ‘Autonomous intelligent agents and the
Roman law of slavery’ in Future Law, eds L Edwards, B Schafer, E Harbinja, 2020