WHO TO SUE? | Your challenges in taking an AI to court!


No longer the stuff of science fiction – Artificial Intelligence (AI) is (almost) here and is here to stay. So, have you ever wondered what you would do if an AI had to injure, defame, infringe or steal from you?

At the Mobile World Congress recently held in Barcelona the topic of Artificial Intelligence (AI) was debated amongst many of the industry’s top thinkers. The prevailing message was that the sheer pace of development of AI is leaving policymakers behind. The development, use and liability of AI is proving to be a legislative minefield.

AI is being incorporated in almost every new technological product, often unnecessarily. Many technology companies are employing the term AI to appear cutting edge. It is critical to define what AI actually is, before even debating the legal liability of AI.

There are numerous definitions of AI, the most widely accepted being that AI is a collaboration of specialised computer software and hardware that together perform tasks usually requiring human intelligence, such as visual perception, speech recognition and decision making.

AI utilises ‘machine learning’ which, in turn, relies on ‘big data’ to learn. Machine learning is where a computer can learn from, and make decisions based on, previous outcomes without being explicitly programmed. Big data is what it sounds like, extremely large data sets which may be analysed computationally.

The reason AI has become so prominent is that we now have the raw computational power for computers to analyse big data faster and more efficiently than ever before. A significant development was the realisation that linearly processing big data, i.e. one piece of data after the next, is hugely inefficient. Consequently, AI using machine learning to process big data, analyses data in a more natural manner in a similar fashion to the manner in which our own brains analyse data.

Insofar as legal liability is concerned, there are two areas of liability which need to be considered: criminal liability; and, civil liability.

Criminal liability is based on criminal law and arises between a person and the state. In order for a person to be found criminally liable for a criminal act or omission, they are required to have both a wrongful mind (intent) and to have committed a wrongful act.

Civil liability, on the other hand, is based on the laws of delict and arises between two or more persons. Delictual liability requires there to be conduct, which is wrongful, causes harm to another and the blameworthiness of which may be attributed to a person by way of conduct or negligence. Civil liability is determined by judging the actions of a person against ‘the standard of the reasonable man’ in order to determine if the person acted reasonably, or failed to act when they ought to have reasonably done so.

Where we are now

AI is currently still relatively rudimentary but is starting to reach the stage where its actions or omissions may lead to some form of legal liability.

Audi, for example, has implemented in their new flagship model, the A8, level 3 autonomous driving.  This means that in traffic the A8 may navigate and drive itself with practically no human involvement, it being piloted by Audi’s on-board AI systems.

When launched, however, Audi cannot distribute the A8 with full level 3 autonomous driving enabled as there are two fundamental questions unanswered by our law, namely:

1) How should the AI

The full article is reserved for our subscribers!

Read this article by Brendon Ambrose, Associate Attorney, Spoor and Fisher, as well as a host of other topical management articles written by professionals, consultants and academics in the April/May 2018 edition of BusinessBrief.

VIEW our subscription options


Questions or problems?

admin@bbrief.co.za | +27 (0)11 788 0880 |



Please enter your comment!
Please enter your name here

For security, use of Google's reCAPTCHA service is required which is subject to the Google Privacy Policy and Terms of Use.

I agree to these terms.