Reco Warns of Growing AI Agent Security Threat at Black Hat USA and DEF CON 34

Reco, an AI and agent ecosystem security company, will present two sessions at Black Hat USA and DEF CON 34 on the growing concern of AI agent security exposure. The talks will focus on the 'security debt' that arises when AI agents become embedded in enterprise applications and workflows without proper governance.

The Reco team will showcase how agents can inherit permissions, use OAuth grants, trigger actions, and expose data through trusted business systems used by organizations. At Black Hat USA, Shir Grinfeld, Head of Product Growth at Reco, will present a three-phase framework for reducing the exposure window created by AI agents.

The framework addresses how organizations can identify misconfigured apps, unenforced SSO policies, and over-privileged tokens before agents find them; detect and respond to machine-speed OAuth probing during an incident; and preserve access trails for post-incident remediation and executive reporting. Reco experts will be available at Black Hat USA booth 1644 to discuss the risks of AI agent security exposure.

At DEF CON 34, Reco researchers Nitay Bachrach and Cynthia Ardman will co-lead a hands-on workshop on Salesforce Experience Sites, one of the most under-tested attack surfaces in enterprise applications. The session is designed for pentesters and red teamers who need a practical methodology for assessing Salesforce sites that standard web testing often misses.