Bonfy.AI CEO Highlights Data-Layer Risk in AI Agent Security

Gidi Cohen, CEO of Bonfy.AI, identifies data-layer risk as the most pressing gap in AI agent security, highlighting that autonomous AI agents operate across systems with no visibility into the data they access, combine, or expose. The industry currently focuses on prompt injection and model behavior, while the deeper threat lies in uncontrolled data movement by these agents. Bonfy.AI addresses this issue through three key areas, though specific technical details or methodologies are not disclosed. The discussion is part of an interview published by Help Net Security on March 23, 2026. No CVEs, numerical data, or explicit impacts are mentioned in the provided content.