What complicates determining liability for custom-made AI systems?

Determining liability for custom-made AI systems is complicated primarily because of the blending of manufacturer knowledge and client specifications. In these scenarios, the responsibility for the system's behavior can be shared between the developer (the manufacturer) and the user (the client). The manufacturer has expertise in building and developing the AI system, incorporating their knowledge into its design. However, clients often specify certain features or parameters that can influence how the system operates.

This dual input creates ambiguity about who is ultimately responsible when the AI system fails or causes harm. For example, if the system behaves unexpectedly or results in negative outcomes, it may be difficult to ascertain whether the fault lies in the manufacturer's design choices, which are informed by their technical understanding, or in the client's specifications that directed the system's development. As a result, this intertwined relationship complicates the establishment of clear legal accountability and the assignment of liability.

In contrast, the other options do not address the nuances of custom-made systems. A standard tort liability framework may not apply straightforwardly due to the unique characteristics of AI. The absence of legal personality does affect how AI systems are treated under the law, but it does not directly complicate liability determination in custom-made cases. Lastly, suggesting that performance is solely based