A Small Minority of AI Users Pose a Huge Security Risk, New Research Reveals
A disturbing trend has emerged in the world of artificial intelligence (AI) security, where only 5% of users are responsible for nearly all cybersecurity breaches. These individuals, often referred to as “superusers,” have inadvertently created backdoors into their organizations’ systems through careless use of AI-powered tools and services.
The issue arises from a phenomenon known as “identity exposure.” This occurs when an individual’s digital identity is compromised or leaked, allowing attackers to assume that person’s privileges within the system. In the context of AI, this can have catastrophic consequences. When an attacker gains access to a superuser’s account, they can exploit their elevated permissions to map cross-domain privilege escalation and create active attack paths.
This means that the attacker can navigate through the organization’s network, bypassing security controls and reaching sensitive areas without being detected. The most vulnerable point in this process is often at “key choke points,” where data flows between different systems or domains. If an attacker can compromise a superuser’s account and position themselves at these critical junctures, they can sever breach routes, making it nearly impossible for security teams to contain the damage.
The implications of identity exposure are far-reaching, affecting not only individual organizations but also the broader AI ecosystem as a whole. As more businesses adopt AI-powered solutions, the risk of superusers creating vulnerabilities increases exponentially. Moreover, this issue highlights the need for better education and awareness about AI security best practices among users.
While most AI users operate within secure parameters, these superusers pose an outsized threat due to their elevated privileges and careless behavior. To mitigate this risk, organizations must implement stricter access controls, monitor user activity more closely, and provide regular security training to all employees – especially those with sensitive roles or access to critical systems.
For individuals working with AI tools, the takeaway is clear: be mindful of your online identity and take steps to protect it. Use strong, unique passwords, enable two-factor authentication whenever possible, and keep your software and operating system up-to-date to minimize vulnerabilities. By doing so, you can significantly reduce the risk of becoming a superuser and inadvertently creating backdoors for attackers.
Source: The Hacker News — 2026-08-24