ทันเอไอ

Global AI & tech news, in your language · every story source-checked

🌐Follow
← Back to news
AI Models & ResearchVerified

"Non-Human Identities" from AI Agents Emerge as New Vulnerability for Organizations

AI Agents are generating a vast number of "non-human identities" that traditional human-centric security systems cannot manage, creating new vulnerabilities for hackers.

📅 12 Jul 2026, 04:22
"Non-Human Identities" from AI Agents Emerge as New Vulnerability for Organizations

Cybersecurity firm Netwrix published an analysis (as a sponsored article on BleepingComputer) highlighting that the proliferation of AI Agents, or autonomous AI programs, presents a major challenge to organizational security. These Agents function as "non-human employees" needing identities and data access rights. Yet, most security systems are designed exclusively for human users, leading to an "Identity Security Gap" that is hard to observe and control.

Data from the industry group Non-Human Identity Management Group (NHIMG) indicates that in many organizations, 'machine identities' already outnumber human users by 50 to 1. Furthermore, Netwrix's "2026 Data and Identity Security Report" found that organizations with a significant increase in identities due to AI adoption had a breach rate as high as 43%, emphasizing the heightened risk.

Such vulnerabilities have already occurred, exemplified by the hacker group UNC6395, which leveraged OAuth tokens (digital keys for inter-app identity verification) from the Salesloft Drift application to infiltrate Salesforce systems of multiple companies. This highlights how connections between applications and AI Agents can serve as backdoors for malicious actors.

This trend is further corroborated by other sources; analysis firm Gartner predicts that by 2029, over 50% of successful AI Agent attacks will exploit access control weaknesses. This indicates that traditional Identity and Access Management (IAM) frameworks are no longer adequate for the AI era.

Why it matters
As Thai organizations increasingly adopt AI Agents, the "non-human identity" vulnerability becomes an imminent risk, potentially leading to the leakage of critical company data and customer personal information.
#AI Agent#Cybersecurity#IAM#Non-human Identity

Related news

โมเดล & วิจัย AI
Bun ย้ายฐานโค้ดจาก Zig สู่ Rust สำเร็จใน 11 วัน ด้วยพลัง AI 'Fable 5' จาก Anthropic
โมเดล & วิจัย AI
Iroh เปิดตัว Mesh LLM: เทคโนโลยีใหม่รวมพลังคอมฯ หลายเครื่อง รันโมเดล AI ยักษ์แบบกระจายศูนย์
โมเดล & วิจัย AI
Notion เปิดตัวแอปแยก 'Notion Agents' ผู้ช่วย AI ส่วนตัวบน iOS เลือกสลับใช้ได้ทั้ง GPT, Claude, Gemini
โมเดล & วิจัย AI
นักวิจัยพบ 'Ghostcommit' เทคนิคใหม่ซ่อนคำสั่งอันตรายในไฟล์ภาพ หลอก AI ขโมยข้อมูลลับ