An AI-powered investment decision-making system lost tens of millions of US dollars of a client's capital, prompting questions about whether the system was fit for purpose, and a legal dispute over who was liable for its losses.

Wanting a fund that would trade with no human intervention so as to remove emotion and bias, MMWWVWM Limited (VWM) asked Monaco-based investment manager Tyndaris SAM to manage its capital using an AI system run on its K1 supercomputer, reckoned to be capable of applying machine learning to real-time news, social media data, and other sources, to predict sentiment in the capital markets.

But VWM quickly racked up an estimated USD 22 million of losses, and asked Tyndaris to suspend its account. Tyndaris responded by claiming approximately USD 3m from VWM in unpaid fees, prompting VWM to counterclaim on the basis that Tyndaris had misrepresentated the capabilities of its system.

The case highlights the challenges in assigning liability for AI-driven decisions, especially in high-stakes financial environments. It underscores the need for clear guidelines on marketing AI capabilities, thorough testing protocols, and robust risk management strategies when deploying AI in investment management.

The outcome of this case, scheduled for trial, may set important legal precedents for future disputes involving AI-powered financial systems.