LSEG

The Workspace agent is described as a reporting layer over existing financial and market data, and World Check Resolve re ranks World Check matches with machine learning. Without the inference, the licensed data platform, analytics and screening content all remain, which is what customers pay for.

Resolve does not make compliance decisions for the customer, Active Intelligence is designed to support analysts, and customers keep control of every final risk decision. Thresholds and audit sampling are unpublished, and for agents built in Microsoft Copilot Studio against LSEG data the oversight controls belong to Microsoft.

Record elements behind each match score are shown in Resolve, which supports case notes and audit evidence for supervisory expectations on AI. The test method, data set and validation package behind its headline result are unpublished, and no supervisory framework is named.

Testing of World Check Resolve cut false positives by an average of 74 percent with no reduction in recall, by LSEG's account. No customer is named with an outcome for Resolve or the Workspace agent.

Whether customer queries, connected data or screening results train models, and how long they are kept, is not stated for Workspace AI or Active Intelligence. Responsible and secure AI appears in partnership language without a term a buyer could rely on.

Developer pages point to the group's standard privacy statement, with no processing terms, retention or isolation published for the generative features or screening data sent to Resolve. A planned phase letting firms combine their own data with LSEG datasets in the Microsoft environment makes that gap concrete.

An information security white paper describes alignment with multiple frameworks and supplier security requirements but names no certification, audit type or trust portal. SOC reports and ISO certification for specific products would be expected from a group of this scale.

Group authorizations for an exchange, clearing house and benchmark administrator do not attach to a research platform, a screening database or their AI features. For the Data and Analytics division, no regulatory position is stated, usual for a data supplier.

Match re ranking decides which names an analyst sees first, and name matching performs unevenly across scripts and transliterations. Fairness testing and a governance framework are unpublished, and whether the false positive reduction holds across name origins is unknown.

A person wrongly matched has no route to LSEG, and when licensed content feeds a customer built agent on Microsoft's platform, responsibility splits three ways. Audit records and score rationale let an institution contest a finding, but LSEG publishes no allocation of loss for screening or agent errors.

Microsoft's AI and Copilot Studio power Workspace, Amazon Bedrock runs Active Intelligence and Resolve, and Babel Street Match powers World Check One's name matching. Model families and versions are not mapped to features, and no subprocessor list is published.

World Check Customer Risk Screener is documented on the public developer portal as a Salesforce Lightning integration using flows or Apex with date stamped audit records. World Check One also ships by API, with data reaching Microsoft 365 through a protocol server and datasets reaching Rogo.

Data distribution can run on site, hosted or in public cloud, but no region, residency or tenancy position is published for the AI features. Where Resolve processes screening data, and whether European data stays in Europe, is open.

Neither Workspace nor the World Check products carry a published price, tier or unit of charge. Whether the Workspace agent or Resolve comes inside existing subscriptions is not stated.

Workspace serves more than 350,000 users across the buy side, sell side, banks and corporates, and Risk Intelligence sells World Check to banks, non bank financial institutions and fintechs. Each line carries its own maintained products.