Large language models regularly hallucinate non-existent web addresses that attackers preemptively register and abuse with phishing pages. Palo Alto Networks Unit 42 documents the “Phantom Squatting” phenomenon for the first time in practice.
Large language models (LLMs) regularly generate invented, non-existent web addresses in their outputs. Security researchers at Palo Alto Networks Unit 42 now observe that attackers deliberately identify, register and populate these hallucinated domains with phishing pages in order to intercept users who trust the AI model.
The business model works because users of AI tools tend to follow the provided links without verifying their legitimacy. When an LLM mentions a domain, it signals trustworthiness, even though the model may have completely invented the address. Attackers monetize this trust gap by acquiring the invented domains and using them for phishing or malware distribution.
For CISOs, this phenomenon represents a new attack class: employees who trust AI tools and click on addresses without verification become a security risk. Half of the threat lies not in technical exploits, but in psychological manipulation through supposedly AI-validated links.
Attackers register domains hallucinated by LLMs and use them for phishing because users trust AI-generated links.
Source: thehackernews.com · Published 1 July 2026
Lumi AI News — AI-assisted curation pursuant to Article 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.2.