Why AI Liability should shift from fault-based theory to Enterprise Liability

Artificial Intelligence (AI) now plays a significant role in individuals’ daily lives, resulting in legal challenges concerning liability for defects in AI systems. Due to AI’s complex characteristics and its autonomy from human control, the application of fault-based liability, focusing on identifying a culpable individual, is no longer appropriate. Traditional fault-based theory requires the claimant to establish that harm was caused intentionally or negligently, emphasising the responsibility of the individual wrongdoer.

This imposes an undue burden on the injured party, who must establish fault before the court, even though the technological complexity of AI makes it exceedingly difficult to demonstrate a clear causal link between the AI’s conduct or decisions and the resulting damage. Accordingly, it is necessary to adopt the theory of enterprise liability, whereby manufacturers of AI systems are held liable for defects regardless of fault. Under this model, liability arises directly from the defective condition of the AI system itself.

The burden on the injured party is limited to proving that the AI was defective and that the defect caused the harm. Enterprise liability focuses on the distribution of risk and the internalisation of production costs within the organisation. This approach enables manufacturers to anticipate and calculate the cost of potential harm in advance, allowing them to distribute the risk among stakeholders or consumers in the market, rather than bearing the entire burden alone.

As such, enterprise liability imposes strict liability on manufacturers, which is a more suitable and effective legal framework for addressing harms caused by AI than traditional fault-based liability.