- AI Agents Pose Insider Risks Due to Unsupervised Access and Lack of Visibility Controls, Report Claims
- 66% of major data losses are due to careless employees or third-party contractors.
- Proofpoint calls for adaptive, behaviorally aware security to protect human and AI activity
As companies rush to deploy Generative Artificial Intelligence (Gen AI) and AI agents, they face the same problems that arise when trying to quickly deploy any new technology: risks of spills, leaks, and data breaches.
A new report from Proofpoint has argued that “agent workspaces” are a new class of insider risk, rivaling in severity even human error. Two in five organizations surveyed said data loss through public or enterprise GenAI tools is a top concern, while more than a third said they are concerned about sensitive data being used in AI training.
AI agents often operate as privileged superusers, the report explains, which only makes matters worse. More than a third (38%) said unsupervised access to data by AI agents is a “critical threat,” and 54% said they lacked sufficient visibility and controls over generation AI tools; In other words, AI agents are being left to their own devices, and this spells trouble.
Testing the limits
“We have entered a new era of data security where insider threats, relentless data growth, and AI-driven changes are testing the limits of traditional defenses,” said Ryan Kalember, chief strategy officer at Proofpoint.
“Fragmented tools and limited visibility leave organizations exposed. The future of data protection depends on unified AI-powered solutions that understand content and context, adapt in real time, and secure information across both human and agent activity.”
Even though AI is a problem, humans are still the weakest link in this cybersecurity chain. Two-thirds (66%) of organizations attributed their major data loss events to “careless employees” and third-party contractors, while 31% cited compromised users. A third (33%) targeted malicious insiders.
To mitigate threats from both humans and AI agents, Proofpoint advises turning toward cybersecurity and behavioral analytics: “just 1% of users are responsible for 76% of data loss events, emphasizing the importance of adaptive, behaviorally aware security strategies,” he says.
Apparently, two-thirds (65%) have already implemented AI-enhanced data security capabilities.