Sumsub has expanded its Know Your Agent framework with software that filters what AI agents can access inside company systems and can require a human liveness check before a sensitive request proceeds. The KYA Module for MCP starts with support for Amazon Bedrock AgentCore Gateway.
MCP, or Model Context Protocol, connects AI agents to tools and data sources. Sumsub places its module between an agent and the information it requests, screening each request and removing personally identifiable or other sensitive information before it reaches the agent. Businesses configure rules governing which files, actions and financial thresholds an agent may access.
An attempted invoice approval above a specified amount, or a request for a restricted document, can trigger additional verification. Sumsub says a real-time liveness check can confirm that an authorized person stands behind the request. A liveness check tests whether a person is physically present rather than relying solely on a submitted photograph or recording.
Organizations can add questionnaires or address verification where the requested action warrants further checks. Rules can take account of data sensitivity, the action being attempted and the departments involved. An activity log records what an agent requested and did, giving the business a record to review alongside its access policies.
The product builds on Sumsub’s agent-verification service, introduced in January to bind automation to a verified human identity. In June, the company added an MCP integration allowing agents to configure and operate its platform. The new module addresses the permissions and monitoring applied when agents connect to company information, with support for additional MCP gateways planned after the initial Amazon integration.
Sumsub says businesses often have limited visibility into what an agent does once it is granted broad access to file storage, communication platforms and financial systems. It cites research finding that 65 percent of organizations experienced at least one security incident caused by an agent operating on their network in the past year, with sensitive data exposure the leading cause.
“This gives businesses a clear record of what their agents are doing and a way to confirm a real person stands behind the actions that matter most,” said Mick Amelishko, AI advocate at Sumsub. The London-based company says it has more than 4,000 clients across financial services, crypto, mobility, trading, marketplaces, education and online gaming.
Sources: Sumsub
–
By the ID Tech Editorial Team
