Global banking cooperative SWIFT plans to launch in 2025, launching fraud detection capabilities using artificial intelligence. The new feature will give financial institutions more accurate information about potential fraudulent activity in real time.
Fraudulent attacks using AI-generated deepfakes impersonating high-ranking executives and AI-generated synthetic identities are only on the rise. According to SWIFT, the global industry estimates the total cost of financial services fraud at $485 billion in 2023.
“Criminals are using increasingly sophisticated tactics to commit financial crimes, and the global financial industry needs to strengthen its defenses to ensure their customers can continue to transact globally with confidence,” said Jerome Piens, chief product officer at SWIFT.
The project relies on the billions of transactions that flow through the SWIFT network each year as training data for the artificial intelligence engine. SWIFT removes identifiers from data to protect privacy and replaces them with pseudonyms.
“This technology will identify suspicious patterns in real time, reducing the risk of fraud and providing a safer banking experience for customers,” added John McHugh, head of CIB operations and compliance at Standard Bank, in a prepared statement.
The new technology is based on the Co-op’s payment monitoring service, which is used by many small and medium-sized financial institutions, and is part of SWIFT’s portfolio of artificial intelligence innovations.
Member banks in Asia Pacific, Europe, the Middle East and North America completed the SWIFT pilot earlier this year.
In addition, the organization is working with large financial institutions to explore other ways to share data between institutions while maintaining data confidentiality. One such technology they are exploring is federated learning, a machine learning technique that trains models on multiple decentralized servers or devices without exposing sensitive data.
Data sharing was also the subject of a panel discussion during the 2024 SWIFT Sibos conference in Beijing. Representatives from Deutsche Bank, Intesa Sanpaolo, UniCredit and SWIFT acknowledged the benefits of data sharing but called on regulators to define a minimum level of data that can be shared to improve fraud detection, Finexta said. Until such regulations are in place, financial institutions will likely not share transaction data.