GSS uses AI tool to speed sanctions screening releases

Global Screening Services has introduced an AI-based cloud review tool developed with Storm Reply. The system cuts preparation time for complex infrastructure releases on its sanctions screening platform from about two days to less than four hours.

The tool, called Cloud Guardian, reviews proposed changes to the cloud infrastructure behind GSS's transaction screening platform before release. It flags changes that may need closer scrutiny, allowing engineers to focus on higher-risk items while keeping final approval with people rather than software.

GSS provides sanctions screening technology to regulated financial institutions. Its platform operates in a market where more than 40 billion cross-border transactions are screened each year for sanctions compliance, often multiple times across different jurisdictions.

This generates a high volume of alerts. Almost 99% of transactions flagged for further investigation turn out to be false positives, according to the company, adding cost and delay to compliance operations.

The new review process targets the infrastructure layer rather than screening decisions themselves. Individual infrastructure reviews that previously took about 30 minutes can now be completed in roughly three minutes, with engineers focusing on issues identified by the AI instead of manually checking every line of a proposed change.

The approach is intended to help GSS update the platform more quickly while maintaining oversight. That matters for sanctions screening systems, which must keep pace with frequent regulatory and operational change.

Recent concerns about sanctions controls have increased pressure on financial firms and service providers to improve their processes. The Financial Conduct Authority has identified weaknesses including outdated lists, poor screening configuration and delays in updates.

GSS raised more than USD $47 million, or about GBP £37 million, in a Series A2 funding round in 2024. It has said its broader model is designed to provide a cloud-based, centralised screening service that reduces repeated checks, false positives and investigation times.

How it was built

Storm Reply worked with GSS platform engineers to build Cloud Guardian as a cloud-native system on Amazon Web Services. The project included testing how the AI should be prompted, selecting a model that could produce reliable analysis at an acceptable operating cost, and linking the software into the existing release process.

Rather than replacing existing infrastructure tools, the system has been added to GSS's current workflow. This has avoided the need for a separate review process and helped engineers build the knowledge needed to maintain the tool themselves, according to the company.

Adam Clarke, Chief Technology Officer at GSS, said: "This development means we can make changes to the infrastructure supporting our sanctions screening service much more efficiently, while retaining the level of control and trust needed for technology used by regulated financial institutions. Sanctions screening has to operate in an environment that is continually changing. We want the platform underpinning it to evolve just as quickly, so improvements can be introduced without creating unnecessary operational friction or risk."

The project reflects a narrower use of AI than some automation efforts in financial services. Here, the software is used to interpret and prioritise technical changes before release, while accountability for production decisions remains with engineering staff.

Storm Reply said this distinction was central to the design. The aim was to reduce the manual effort involved in assessing infrastructure changes without removing the controls expected around systems used by regulated financial institutions.

Rachel Grunwerg, Partner at Storm Reply, said: "The challenge was to help GSS move faster without weakening the controls around a highly important financial services platform. The biggest opportunity for AI here was not to automate the final decision, but to reduce the amount of manual interpretation engineers had to do before they could make that decision. That is where we see significant potential for AI across financial services: helping people work through complex technical information faster, while keeping expertise, governance and accountability firmly with those responsible."