BSI Group Workshop - Ensuring trust in Agentic AI - certification of AI agents

Agentic AI, valued at ~ $9–12B globally, is the fastest-growing segment, expanding at CAGR > 40%. The shift from conversational chatbots to autonomous agents capable of independent planning, tool usage, and complex reasoning without human intervention is being driven by an enterprise push to automate multi-step workflows. In terms of enterprise penetration, 45% of organizations operating AI agents at scale by 2030 (Source: IDC), with 62% currently experimenting with AI agents (Source: McKinsey). While generative AI automates the creation of complex text, images, and video based on human language interaction, AI agents go further, acting and making decisions in a way a human might. 

Enterprises are aware of AI risk but can’t assess it confidently. Leading to:

• Protracted procurement processes
• Security and legal discussions diverted to answer security questionnaires
• Technical discussions missing the point.

Certification is part of the solution to the problem of enterprise trust and liability when adopting agentic AI. Autonomous agents actively plan, use external tools, and access data (rather than just answering prompts), they create massive risks like tool misuse, data leakage, and hallucinations. The way to resolve this is certification that addresses this through continuous adversarial stress-testing rather than just reviewing policy. 

AIUC is a standard for AI agent security, safety and reliability, designed to assess agentic AI systems against a structured set of controls and technical safeguards. AIUC certification evaluates specific agent deployments to determine whether they are operating responsibly with appropriate controls, accountability structures, and oversight mechanisms in place. 

AIUC certification covers the full AI lifecycle from how a use case is defined and documented, through how risks are identified and managed, to how the system is monitored once it’s in production. For agentic systems, this means assessing not just what the AI is designed to do, but how it behaves autonomously across a range of real-world conditions. 

The key learnings for the delegates that attend the workshop include:

• Understand how agentic AI differs from traditional generative AI systems.
• Identify the unique safety, security, and governance risks introduced by autonomous AI agents.
• Learn why traditional vendor due diligence processes struggle to assess agentic AI systems effectively.
• Understand how independent assurance and certification can support trust in enterprise AI adoption.
• Explore how AIUC evaluates agentic systems across governance, technical controls, and operational monitoring.