Who gets it wrong: The dispute lifecycle of AI claims – from product design to group claims and arbitration

As AI systems become embedded in global commerce, disputes increasingly arise not only from how these technologies are used, but from the choices made long before deployment - data sourcing, model training, product design, and governance. This session traces the full lifecycle of modern AI disputes, from early IP‑ and design‑linked risks, through operational failures and mass‑harm scenarios, to cross‑border arbitration where evidence, confidentiality and procedural fairness come sharply into focus.
Drawing on real developments - from challenges around AI‑generated outputs and training data rights (as illustrated by litigation involving synthetic watermark outputs) to cases where AI‑driven decisions have triggered commercial claims, and to collective action trends shaped by UK procedural limits and EU reforms, the panel will explore how and why AI "can get it wrong"  / the use of AI can “go wrong,” and what parties must do to prevent disputes arising/ escalating.