About the Public Service Catalog
F5 Workforce AI Security continuously finds and catalogs public generative AI services on the internet. When Workforce AI Security finds a service, it collects data about the service and adds it to the Public Services list. You can then allow or block user access to each service.
Data for each service
To see the full list, in GOVERN, go to Public Services. You can also go to https://admin.surepath.ai/config/public-services.
Status
Shows whether the service is available and whether Workforce AI Security supports it.
AVAILABLE: The service is available on the internet, and Workforce AI Security supports it.
DEPRECATED: The service isn't reliably available, or its vendor plans to retire it. Workforce AI Security gives limited support and plans to remove the service.
Risk Level
Shows the risk level of the service: LOW, MEDIUM, or HIGH. Workforce AI Security sets the level from three factors:
Use of user data to train models, shown in Trains on User Data
No content moderation, shown in Unmoderated
Risk level of the country where the company that runs the service is based, shown in Country.
When the value of a factor isn't known, Workforce AI Security counts it as a risk. When these factors don't show a known risk, a reviewer can set the risk level of a service manually.
The risk level of the service is one part of the risk assessment for each prompt. For more about risk assessment, see Risk assessment overview.
Risk Confidence
Shows how much evidence supports the risk level. Risk Confidence doesn't change the risk level.
LOW: Little evidence supports the risk level.
MEDIUM: Some evidence supports the risk level.
HIGH: Strong evidence supports the risk level.
Trains on User Data
Shows whether the vendor uses user input to train or fine-tune its models.
No: The vendor doesn't use user input to train its models.
Yes: The vendor can use user input to train or improve its models.
n/a: The value isn't known.
Some services have subscription plans, often paid, that stop the use of user input for training.
Unmoderated
Shows whether the service filters user input or generated output. When an upstream model provider filters the content, the service counts as moderated.
No: The service moderates input or output.
Yes: The service doesn't moderate input or output.
n/a: The value isn't known.
Country
Shows the country where the company that runs the service is based. This isn't always the country where the service is hosted. The risk level of this country is one of the risk factors for the service.
Intercept Support
Shows the types of traffic that Workforce AI Security can intercept for the service: Request, Response, License, File Upload, and MCP (Model Context Protocol).
With request intercept, Workforce AI Security can capture, log, and change the prompt before the service receives it. Workforce AI Security needs request intercept to stop sensitive data from going to the service.
With response intercept, Workforce AI Security can capture and log the response that the service sends to the user.
You can allow or block access to all services in the catalog, whatever their intercept support. To do this, use the Default Policy and Group Policies.
Company
The company that develops, maintains, and runs the service.
Description
An overview of the service. It includes the purpose of the service, its approved uses, its limits, and other data that helps you govern its use.
Website
The web address of the service.