Cyber Risks · The Hacker News

How AI Hallucinations Are Creating Real Security Risks

Summary: A 2025 Artificial Analysis AA-Omniscience benchmark of 40 AI models found that all but four were more likely to give a confident, incorrect answer than a correct one on difficult questions. The article describes how such outputs, which cite nonexistent sources and present fabricated data with the same conviction as trusted information, can drive security decisions and feed automated systems that trigger operational actions.

Key Takeaway: The article frames hallucinations as an access problem: incidents arise when AI systems hold enough privilege to act on incorrect guidance, such as deleting sensitive files or disabling firewall rules, or when a human trusts the output without verifying it.

Read the original at The Hacker News ↗

Topics
OT / ICS & Critical Infrastructure
Related
Siemens RUGGEDCOM APE1808 Cyber Risks · Nov 12, 2026