Employees' personal AI agents pose corporate security risks

Fast Company analysis of risks from personal AI agents (Meta's Muse, OpenAI's Dots, Grok Bot, Instinct) in enterprise settings. 82% of organizations have unknown AI agents running in their systems, 65% suffered agent-related security incidents in the past year, and 52% of employees use unapproved AI tools.

• 82% of organizations don't know about AI agents running in their systems
• 65% of companies suffered agent-related security incidents
• 52% of employees use unapproved AI tools
• Meta's Muse hit 2.8 million installs in 12 days