AI agents risk repeating 1974 Herstatt settlement failure in multi-leg trades
The 1974 collapse of Germany's Bankhaus Herstatt exposed a critical flaw in financial settlement: one party paid while the other defaulted before completing their leg of the trade, a risk now known as Herstatt or settlement risk. Traditional finance took 28 years to address this by building CLS, a central institution that ensures both legs of a foreign exchange trade settle simultaneously or not at all. A recent study by Keyrock, Coinbase, and Tempo recorded 176 million machine-to-machine transactions worth $73 million between May 2025 and April 2026, signalling the rapid growth of AI agent-driven commerce. Unlike simple payments, multi-leg agent trades — where one transaction depends on the completion of a prior one — face an elevated risk of partial settlement, especially since AI agents frequently transact with unknown counterparties and can execute thousands of sequences before any human oversight kicks in. Analysts warn that without atomic settlement infrastructure designed for agent workflows, the AI economy could be exposed to the same systemic failure mode that destabilised global banking half a century ago.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in