FS giants test blockchain and AI to tackle $58bn corporate actions problem #
A host of banking and capital market players - including Swift, the DTCC and UBS - have closed out the second stage of a Chainlink-led project to use AI and blockchain technology to standardise and streamline corporate actions processing.
Corporate actions processing costs the global financial industry an estimated $58 billion annually, with costs increasing by 10% year over year and automation rates falling below 40%, claims Chainlink.
According to Citi’s 2025 Asset Servicing report, the average corporate action event now touches more than 110,000 firm interactions and costs $34 million to process, with 75% of market participants still relying on manual data revalidation.
Last year, Chainlink, Swift, Euroclear, and six financial institutions joined forces to demonstrated that large language models — including OpenAI’s GPT series, Google’s Gemini series, and Anthropic’s Claude series — could extract structured data from unstructured corporate action announcements and publish it as unified golden records onchain — a single source of truth that all participants can easily access, verify, and build upon.
For phase 2 of the project, 24 organisations - including ANZ, DBS Bank and TMX - introduced a production-grade deployment by making substantial improvements to the speed, reach, and accessibility of corporate actions data in a move the partners claims could save the the global financial system tens of billions of dollars a year.
Testing saw the Chainlink Runtime Environment orchestrate the validation of multiple AI model outputs and transform the confirmed results into ISO 20022-compliant messages, which were then transmitted to the Swift Network.
In parallel, Chainlink Cross-Chain Interoperability Protocol distributed these same confirmed records across DTCC’s blockchain ecosystem and additional public and private blockchain environments, enabling simultaneous access across traditional infrastructure and blockchain-based platforms.
New institutionally-designated roles for data attestors and contributors were also introduced to cryptographically attest to data accuracy and contribute to any missing data fields, creating a verifiable chain of custody across the lifecycle of each corporate action.
Throughout testing, the system achieved nearly 100% data consensus agreement among AI models across all evaluated corporate actions.
The architecture also demonstrated support for multilingual processing across disclosures written in non-English languages, such as Spanish and Chinese.
Chainlink says the system unlocks a unified golden record for corporate actions: an attested, real-time source of truth that can be accessed simultaneously by smart contracts, custodians, and post-trade systems.
It also enables tokenised equities, an increasingly adopted category of tokenized assets, to reference the same confirmed records across public and private blockchains, laying the groundwork for better synchronisation and increased automation across onchain markets.
The group now plans to extend the workflow to support more complex corporate actions, like stock splits, expanding global coverage through support for additional jurisdictions and currencies, and introducing stronger privacy and governance controls.
Nigel Dobson, banking services lead, institutional, ANZ, says: “The Chainlink-led initiative demonstrates how blockchain can deliver attested, real-time corporate actions data across jurisdictions and platforms—reducing operational risk, enabling smart contract automation, and laying the foundation for scalable digital asset servicing.”