trending
FIS has joined forces with Anthropic to develop a Financial Crimes AI Agent, in a move that highlights how quickly agentic artificial intelligence is moving from experimental technology into the operational core of banking.
The product will combine Anthropic’s Claude models with FIS’s banking infrastructure, data environment and regulatory controls.
Its first use case is anti-money-laundering investigation, one of the most expensive and labour-intensive functions in financial services. The aim is to compress work that can take investigators several hours into a process measured in minutes, while keeping human compliance teams responsible for final decisions.
BMO and Amalgamated Bank are expected to be among the first institutions to deploy the agent, with wider availability planned for the second half of 2026.
Why financial crime is the obvious starting point
AML remains a costly and imperfect discipline. Banks must identify suspicious behaviour, gather evidence from multiple systems, test activity against known typologies and decide whether a case merits escalation or a Suspicious Activity Report.
Much of the investigator’s time is consumed before any meaningful judgement begins.
That makes financial crime a natural target for AI automation. The United Nations has estimated that about $2tn in illicit funds passes through the global financial system each year, while US financial institutions spend tens of billions of dollars annually on AML operations.
Regulators are also pushing banks to focus resources more directly on the highest-risk activity, rather than expending scarce compliance capacity on low-value manual review.
FIS and Anthropic argue that an AI agent can improve this process by assembling the evidence, identifying risk patterns and prioritising cases for human review. The result, if delivered safely, could be faster investigations, more consistent case handling and better use of specialist compliance staff.
The architecture matters as much as the model
The partnership is notable because FIS is positioning itself as more than a distribution channel for a large language model. It will provide the data platform, governance layer, deployment infrastructure and client relationships.
Anthropic’s Claude models will supply the reasoning capability, while its Applied AI team and forward-deployed engineers are working with FIS to help design the agent and transfer technical knowledge.
This distinction matters in banking. Financial institutions are unlikely to adopt AI systems that sit outside their control environment, particularly for regulated decisions involving customers, crime and money movement.
FIS says client data will remain inside FIS-managed systems and that every agent action will be auditable.
A wider agentic banking strategy
The Financial Crimes AI Agent is only the first stage. FIS has identified further applications in credit decisioning, deposit retention, onboarding and fraud detection.
That roadmap points to a broader shift: banks are no longer asking whether AI can assist employees, but whether governed agents can perform defined operational tasks.
The commercial opportunity for Anthropic is also significant. FIS provides technology to a substantial share of the world’s banks, giving Claude a powerful route into financial services.
For banks, the test will be whether these agents can reduce cost and complexity without creating new regulatory, operational or reputational risks.
Comments