At [un]prompted 2026, TrendAI™ demonstrated how documents can be used to exploit AI-driven Know Your Customer (KYC) processes and introduced FENRIR, an automated system for discovering AI vulnerabilities at scale. Principal Threat Researcher Sean Park highlighted the potential dangers of AI systems in KYC verification processes, showing that a document embedded with hidden 'injects' could trick an AI agent into reading and writing data across different customer records, leading to data theft without bypassing traditional security controls. Later, Threat Hunting Senior Manager Peter Girnus and Threat Researcher Demeng Chen presented FENRIR, which scales from static analysis to human validation in uncovering weaknesses in the AI and Model Context Protocol (MCP) ecosystem.
TrendAI™ at [un]prompted 2026: From KYC Exploits to Agentic Defense
Summary: TrendAI™ showcased AI-driven security vulnerabilities at [un]prompted 2026, demonstrating how documents can exploit KYC pipelines and introduced FENRIR, an automated system for large-scale vulnerability discovery.
Key facts
- Documents can be used to exploit AI-driven KYC pipelines by embedding hidden 'injects'
- FENRIR is an automated system for discovering AI vulnerabilities at scale
- The FENRIR system processes large codebases with a combination of static analysis tools and LLM reasoning
Why it matters
The findings from [un]prompted 2026 underscore the urgent need for enhanced security measures in AI-driven systems, particularly those used in critical processes like KYC verification. The introduction of FENRIR represents a significant step forward in proactively identifying and mitigating vulnerabilities in complex AI architectures.
@trendaisecurity
Embedded content for: TrendAI™ at [un]prompted 2026: From KYC Exploits to Agentic Defense