Ensuring Trust in AI Agents with OneTrust and Strategy

Intelligent agents are reshaping how organizations operate, but they also introduce new challenges regarding governance, transparency, and accountability.

Trust remains a major barrier: recent industry research shows that only 46% of people globally are willing to trust AI systems. Closing this trust gap is essential as organizations move toward more autonomous, agent-driven AI.

Enterprise AI World, in partnership with DBTA, held a round table webinar, Fostering Trust in AI Systems: Governance for the Age of Intelligent Agents, featuring Tohsheen Bazaz principal product manager at OneTrust and Aidan Reilly, product management director at Strategy, who explored how leading organizations are building governance frameworks that make AI systems trustworthy, auditable, and enterprise-ready.

There is an AI governance blind spot, Bazaz said, with the gap between use case approval versus what was built widening.

According to Bazaz, implementing AI governance at the beginning of a project entails asking and doing the following:

• Do I have information about everything being planned? Leverage AI governance solution to embed collection at source.
• Do I have visibility into what runs in product, staging, development? Integrate data in use, internal/external, consent, model use, and model risk within governance.
• Are my decisions enforceable or create a path of remediation? Governance decisions must be respected and enforceable.
• Do I have tools to assess org risk? What is our gap between being compliant with laws? Leverage curated framework stacks and measurements from AI governance solutions.
• Which technical signals should I worry about to capture governance checks? Guardrail settings, evaluation score, usage metrics, interventions, PII detection, and token use.

Discovering AI with depth helps govern AI based on run time behavior, Bazaz explained. The OneTrust value proposition for AI governance streamlines AI governance for enhanced compliance and innovation.

Reilly noted that the modern data stack consists of the application layer, semantic layer, data platform layer, and the ETL layer.

Capabilities of the semantic layer include:

• Rich semantics and business definitions
• In-memory processing and virtualization
• Consumer-grade AI-powered modeling
• Optimized universal access
• Governance and single pane of glass
• Portable and independent

For AI projects to be successful, there needs to be a top-down organizational buy-in to AI transformation, Reilly stressed.

There should be high usage of MCP and API services for connectivity; governance and security enforced and limited data access; centralized auditing of AI write actions; and reuse of existing investments especially in security/access control.

“Treat AI Agents just as if they were another employee,” Reilly said.

With Strategy Mosaic Sentinel, users get a platform that delivers an omnipotent view of your data landscape, risk management, audit and compliance, performance optimization, and cost intelligence.

For the full webinar, featuring a more in-depth discussion, Q&A, and more, you can view an archived version of the webinar here.