Building Trust and Security for AI Agents

Jetstream CEO Raj Rajamani discusses how organizations can manage risks as AI agents become part of business workflows. The conversation outlines a trust-based approach to agent security and considers how international talent contributes to cybersecurity work.

■ AI agent security
- Agent ownership, defined goals, permitted relationships, sandboxing, monitoring, human approval, and emergency shutdowns
- The practical challenge of defining what agents may and may not do
■ Talent and cybersecurity
- Rajamani’s perspective on international hiring and Jetstream’s distributed workforce

For security leaders and organizations deploying AI agents, this discussion offers a high-level view of safeguards to consider. It does not provide a technical implementation guide; viewers can use the topics raised to review agent oversight, permissions, and escalation plans in their own settings.