Who Owns AI Inference Legal Liabilities?

Before we can discuss who owns Artificial Intelligence (AI) inference legal liabilities, we should discuss what, exactly, AI inference is.??In simple terms, it means that an AI engine is applied to derive new insights or knowledge by providing it with new data set(s). It is not important whether the AI engine is in the cloud or at an on-premises Data Center or some combination of both.??

As AI gets used more and more in both professional and personal settings, the inevitable question arises: “Who is liable if AI provides the consumer with inaccurate outcomes which may cause harm to others – the entity who produced the AI engine or the entity who inferred the AI engine?”

There is no simple answer to the question above; to make things more complicated, it also depends on what part of the world the inference took place because, as we all know, laws, regulations, and compliance vary widely across the globe. As an additional layer of complexity, we must also consider the issue/question of Intellectual Property (IP).

Let us envision a scenario in the creative field;??if we take the famous picture of the Mona Lisa and superimpose our own face over it using AI and call it our own art, or if you were to use AI to superimpose your own music over a famous piece of music, the Japanese government would not grant you IP (Intellectual Property) for that (and the European Union is quickly coming to that same position). Laws are not yet well-defined regarding IP in the creative space of art and music.?

Now let us imagine another scenario in the automotive field; let’s say you work for a company which develops autonomous vehicles. As we all know, AI is a cornerstone for autonomous vehicles’ driving capabilities, so you (as an engineer) use some other company’s AI as a part of your autonomous vehicle design, allowing images captured by your vehicle’s cameras to be inferred and to aid the autonomous capabilities of your vehicles. If one of your consumers, driving one of your vehicles, gets into an accident, you and your company may be considered liable – not the manufacturer of the AI engine you incorporated into your vehicle’s design. As you and your company are using AI to “infer” visual images for autonomous driving, who would ultimately be responsible if an accident happens??

The above examples are merely explanatory, and not intended to set any sort of legal precedent. The point here is that the entity inferring is legally liable, not the AI engine producer – but as we mentioned earlier in this post, this would also depend on your geographical area’s laws.

Laws and regulations do continue to evolve, and we are sure this view – and the legal implications of AI inference – will inevitably change in the future.?

This is a joint article by René Aerdts and Hemang Davé. Please note that the views expressed here are ours only, and do not represent Kyndryl’s official positions. We would love to hear how you would approach this topic, or your thoughts on regulation of AI or AI inference legal liabilities. Please feel free to share this article with others in your network.