Swift and Google Cloud Collaborate on AI Model Training

Swift and Google Cloud are collaborating to create a safe, private solution for financial institutions that employ federated learning methods.

Google Cloud stated in a blog post on Tuesday, December 10, that federated learning allows for collaborative artificial intelligence (AI) model training without sacrificing privacy and confidentiality.

According to the post, Swift intends to collaborate with Google Cloud and twelve international financial institutions to launch a sandbox with synthetic data in the first half of 2025.

“This exploration will help the community validate whether federated learning technology can help financial institutions stay one step ahead of bad actors through sharing of fraud labels, and in turn enabling them to provide an enhanced cross-border payments experience to their customers,” Rachel Levi, head of AI at Swift, said in the post.

According to the post, this project would also involve collaboration with technology partners such as Rhino Health and Capgemini. The core federated learning platform will be created and delivered by Rhino Health, while Capgemini will oversee the integration and deployment of the solution.

Bank-specific encrypted data that stays secured throughout the process, a federated client that completes tasks and submits findings for secure aggregation, and a federated learning server that coordinates the clients' collaboration are the components of the solution, according to the post.

According to the post, it additionally incorporates a global fraud-based model, safe aggregation, and a global anomaly-detection trained model.

“Our collaboration with Swift exemplifies the transformative potential of federated learning and confidential computing,” Andrea Gallego, managing director, global go-to-market incubation at Google Cloud, said in the post. “By enabling secure collaboration and knowledge sharing without compromising data privacy, we are fostering a safer and more resilient financial ecosystem for everyone.”

In October, Swift said that it was strengthening its AI-powered fraud detection capabilities and that it will start providing this service to the payments industry in January.

Using pseudonymized data from the billions of transactions flowing over the Swift network, the new feature builds upon Swift's Payment Controls Service by identifying questionable transactions and enabling real-time response.

In order to ascertain how AI "can solve cross-industry challenges," Swift stated at the time that the endeavor is a component of its broader collaboration with its network of over 11,500 banks and financial institutions.