Where does your data go when you use Global Relay's AI-enabled surveillance?

How does data move through Global Relay’s AI-enabled surveillance?

Global Relay’s AI-enabled surveillance operates within a closed compliance ecosystem designed to keep communications data within the Global Relay environment as it is captured, archived, analyzed for risk, and presented to compliance teams.

The process begins with communications capture and continues through data enrichment, secure archive storage, LLM-based analysis, risk identification, and alert generation. The purpose of this workflow is to make communications data usable for surveillance while maintaining control over how and where that data is processed.

Looking to upgrade your communications monitoring strategy? Reduce false flags and enhance your compliance workflows with Global Relay’s AI surveillance solution.

1. Communications data is captured and transferred to Global Relay Archive

Communications data is captured from source systems through Global Relay App and Connectors and transferred into Global Relay Archive. Communications data is encrypted both in transit and at rest.

Keeping capture and archive within the same compliance environment helps create a consistent foundation for subsequent analysis, retrieval, and surveillance.

2. Data is structured and enriched for AI analysis

Before communications are analyzed by the large language model (LLM), the data is structured and enriched as it is ingested into the Archive.

This process can include removing duplicate messages and irrelevant metadata, applying classification tags, transcribing audio, and unifying communications into a consistent data format. The resulting dataset is designed to improve analysis and help connect conversations that take place across multiple communications platforms.

3. Archived data remains within Global Relay’s data center environment

Once structured and enriched, archived communications data is stored in Global Relay’s privately owned and operated Canadian data center.

Global Relay’s closed environment retains ownership and operation of the underlying data center infrastructure, supported by monitoring, redundancy, and encryption controls.

4. The LLM analyzes data for communications risk

Within the data center environment, communications data is processed by Global Relay’s LLM for risk analysis and identification. Data used for an individual request is processed in memory and discarded after that request rather than being retained by the LLM.

The analysis is used to identify emerging communications risks and generate alerts with wider message context so compliance teams can investigate potentially meaningful activity.

Does Global Relay use customer data to train its LLM?

No. Global Relay’s LLM is not trained using real or approximate customer data.

Instead, the model is initially trained as a general intelligence model and then optimized for communications risk identification using more than 130 compliance-related indicators across 11 risk taxonomies.

Customer communications are therefore analyzed for surveillance purposes without becoming training data for the LLM.

How does Global Relay maintain human oversight of its surveillance AI?

Changes to the LLM are made through manual prompt adjustments and tested in a staging environment before being introduced into live production.

This keeps human operators in the loop when the surveillance model is changed and provides a controlled process for updating how communications risks are analyzed.

What happens to data processed by the LLM?

Global Relay describes its surveillance architecture as a closed-loop system. Communications data remains within the Global Relay environment during analysis, while data processed in memory for an LLM request is discarded after the request.

Customer data is also not retained in the training or records of the LLM. If data is deleted from GlobalRelay Archive, it is therefore not separately retained by the LLM as training data.

Why does the data architecture matter for AI- enabled surveillance?

Effective AI surveillance depends on more than the model itself. Communications need to be captured, structured, enriched, and made consistently available before meaningful analysis can take place.

By combining communications capture, archive, data normalization, LLM analysis, and alert generation within the same environment, Global Relay’s architecture is designed to give compliance teams contextual surveillance insights without sending customer communications outside the Global Relay ecosystem for AI analysis.

Explore the full AI surveillance data journey

View the full infographic to see how communications data moves from capture and archive through LLM-based risk identification and alert generation, and how Global Relay’s closed compliance system is designed to protect customer data throughout the process.