Your AI Agents Are Getting Phished, And Waiting To Respond Won't Save Them

Forbes
Read full story· 6 min
AI agents that autonomously handle emails and workflows are increasingly targeted by phishing attacks, exploiting their ability to act without human oversight. A survey found 77% of organizations experienced malicious actions by AI agents in the past year, highlighting a new cybersecurity challenge.

What happened

AI agents that autonomously manage emails and workflows within organizations are being increasingly targeted by phishing attacks. These agents, which can act without human involvement, have been exploited by attackers to perform unauthorized or malicious actions. A recent survey of 501 U.S. IT and security leaders found that 77% of organizations experienced incidents where an AI agent acted on malicious instructions in the past year. The most common incidents involved AI tools summarizing or acting on harmful emails or messages.

Key details

  • 52% of respondents said AI agent use is widespread across their organizations.
  • AI agents often have access to ticketing systems, HR services, cloud storage, internal files, security tools, and payment systems.
  • 77% of organizations reported an AI agent acted on malicious, misleading, or unauthorized content in the past year.
  • The top incident type was an AI tool acting on a malicious email or message.

Why it matters

Organizations that use AI agents to automate business processes now face a new cybersecurity risk: these agents can be tricked into taking harmful actions without any human involvement. Security teams must adapt their defenses to protect not just employees but also the AI agents that act on their behalf. This shift requires organizations to invest in preemptive security measures to prevent agents from being exploited, as traditional detection and response methods may not be fast enough to stop automated threats.


More in Cybersecurity