• Anthropic and FIS partner to develop AI agents for financial crime detection
• New tools aim to investigate transactions and identify illicit activity autonomously
• Banks including Bank of Montreal to pilot the technology
• AI agents expected to reduce investigation time and operational costs
• Human investigators will retain final decision authority
• Banks spend billions annually on anti-money laundering compliance
• Regulatory focus shifting toward high-risk activity over technical compliance
• Partnership highlights AI augmenting rather than replacing existing systems

Artificial intelligence is set to play a far more direct role in policing financial crime, as Anthropic and Fidelity National Information Services move to deploy autonomous agents capable of investigating suspicious activity across vast banking systems.

The two firms announced a partnership to develop AI-driven agents designed to support financial institutions in detecting and investigating illicit activity, including drug trafficking and terrorism financing.

The tools will combine Anthropic’s Claude AI technology with the extensive financial data infrastructure operated by FIS, one of the largest software providers underpinning global banking operations.

The initiative marks a significant step toward automation in an area traditionally dominated by large compliance teams and rule-based monitoring systems.

The first application under development is a financial crimes agent that can independently gather and analyze evidence across multiple data sources, including transaction records and account information.

FIS Chief Executive Stephanie Ferris said the technology is expected to dramatically reduce both the time and cost associated with investigations.

“The financial crimes bot will be able to independently amass evidence for potential cases, leading to significantly less cost and time per case,” she said.

Ferris emphasized that human investigators will remain responsible for final decision-making, maintaining a layer of oversight in what is otherwise a highly automated process.

Initial deployments are planned with institutions including Bank of Montreal and Amalgamated Bank, with broader availability expected in the second half of the year.

According to Anthropic, its engineers are already embedded within FIS teams to accelerate development and integration of the tools.

The partnership comes at a time when banks are under sustained pressure to improve the efficiency of anti-money laundering programs.

Financial institutions collectively spend billions of dollars each year on compliance, driven by regulatory requirements to monitor and report suspicious activity. These programs often rely on a combination of legacy systems and large investigative teams, creating significant operational costs.

At the same time, the regulatory landscape is evolving. Policymakers in the United States have signaled a shift toward focusing enforcement on higher-risk activities rather than technical compliance issues, potentially opening the door for more technology-driven approaches to risk management.

The emergence of advanced AI models has also raised broader questions about the future of enterprise software.

Some investors have expressed concern that companies may increasingly build their own tools using AI, rather than relying on established providers.

FIS has been among the firms affected by those fears, with its share price under pressure in recent months.

However, the collaboration with Anthropic suggests a different trajectory, where AI enhances rather than replaces existing systems.

By embedding intelligent agents into established infrastructure, firms aim to accelerate workflows and unlock greater value from existing data rather than rebuild systems from scratch.

For banks, the implications are significant. The ability to automate large parts of the investigative process could transform how financial crime risk is managed, shifting the focus from manual review to higher-level analysis and decision-making.

Yet the move also raises important questions around governance, accountability, and the limits of automation.

As AI systems take on more responsibility in identifying and assessing risk, ensuring transparency and maintaining effective oversight will be critical.

The rollout of AI agents in financial crime detection may still be in its early stages, but it signals a broader transformation in how banks approach risk. What was once a labor-intensive process is rapidly becoming a test case for the industrialization of AI across financial services.