Data Governance · 3 min read
Singapore to Build Cross-Bank AI Scam Detection Utility
Singapore's MAS unveiled three AI initiatives at the FinTech Festival: cross-bank scam detection, AI-assisted SME lending, and supervised payment agents. Five banks are already testing.
Key takeaways
- Deputy PM Gan Kim Yong announced three AI initiatives at the Singapore FinTech Festival on November 18
- Five banks are already testing a system that securely combines transaction data for scam detection
- MAS begins a design study for an industry-wide Scam Detection Utility in early 2026 per the announcement timing
- AI payment agents will operate within approved limits, with financial institutions liable for their actions
- The Future of Finance Institute will partner with industry and researchers on practical agent solutions

Five Singapore banks are already testing an AI system that pools transaction data across institutions, and the Monetary Authority of Singapore plans a design study for an industry-wide Scam Detection Utility starting early next year. Deputy Prime Minister Gan Kim Yong, who chairs the MAS, announced the plans at the Singapore FinTech Festival on November 18. Three initiatives, one message: Singapore wants AI running the plumbing of its financial sector, not just its marketing.
Why does cross-bank detection matter?
Because stolen money doesn't respect bank boundaries. "Scams are not confined within the boundary of one bank," Gan said, noting that illicit funds can move through several financial institutions within minutes. Any single bank's transaction view is, by construction, incomplete — the money leaves before the pattern is visible.
The five-bank pilot securely combines transaction data so AI models can spot scam patterns individual banks miss. The MAS design study, expected to begin early next year, will work out how banks can share that information without handing each other raw customer data. The initial scope covers accounts and financial data inside Singapore; authorities may extend cooperation to other countries later, since cross-border scams remain the hard part. That's an honest admission, not a footnote — most scam flows exit the jurisdiction within minutes, so domestic detection alone caps the upside.
What changes for SME lending?
The second initiative targets small and medium-sized enterprises, which routinely struggle to secure loans because of thin credit histories or a lack of collateral. The fix is data breadth: AI models assessing borrowers using bank account information, invoices, payments and government records rather than traditional credit files. The promised outcomes are faster loan decisions and lower borrowing costs. Whether lenders actually price that risk reduction into rates is the part to watch.
Who is liable when an AI agent pays the wrong bill?
The third initiative is the most interesting risk question. Singapore will test AI agents that make payments and purchases on behalf of individuals and companies, operating within approved limits. The liability assignment is explicit: financial institutions remain responsible for the actions of their agents.
"However, greater autonomy must come with clear responsibility and accountability," Gan said. The Future of Finance Institute will work with industry partners and researchers to develop practical solutions — governance frameworks, presumably, rather than proofs of concept.
What should data teams take from this?
The initiatives sit inside Singapore's broader push to stay a global financial-technology center, so expect regulatory scaffolding to follow the pilots. Three practical signals:
- Federated or privacy-preserving data sharing is moving from vendor decks to bank pilots. Five banks testing combined transaction analysis is a live reference architecture, not a white paper.
- Alternative-data underwriting is getting official air cover. Teams holding invoice and payment data at fintechs and banks should expect inbound integration requests.
- Agent liability lands on institutions. Anyone building agentic payment features should design audit trails and spending limits now, before a MAS framework makes them mandatory.
Singapore's regulators have a habit of piloting first and legislating second. Data teams elsewhere can treat these three programs as a preview of the questions their own regulators will ask within a couple of years. The scam utility alone answers one of them: the useful data isn't in your bank. It's in all of them.
Google News: Business Intelligence (Source)
More from Grace Kim
Show full bio
Market editor covering companies, deals and strategy in Analytics Digest is a professional daily at Analytics Digest.
18 articles