Agentic AI has rapidly emerged as a trend topic, with its adoption gaining increasing traction. Analysts forecast a substantial 33% growth by 2028, combining AI-based assistants responding to user prompts with completely autonomous agent-based workflows. However, the technology’s development is still in its earliest stages, with protocols such as MCP and A2A still under development and undergoing constant revision.
Between the enthusiasm with Agentic AI Adoption and the complexity of implementing it lies the concerns about agent’s authentication and authorization, privacy, data protection, risk & compliance, and the overall impact that agents will bring to the existing organization’ security controls.
For instance, while MCP offers a promising interoperability standard, its authorization model based on OAuth2.1 leaves gap. NHI lifecycle and governance capabilities are essential, but it remains unclear how to govern hundreds of thousands of agents, especially when MCP/OAuth2.1 proposes Dynamic Client Registration as the agent registration solution. These are just two examples of potential solution gaps.
In this presentation, we will explore the latest advancements in Agentic AI technologies and delve into the risks and security challenges organizations will encounter when deploying Agentic AI at scale. We will also discuss the required Cyber and Identity capabilities that organizations must have in place before embarking on the Agentic AI journey.
