Attackers are exploiting a phenomenon called ‘phantom squatting,’ where large language models (LLMs) hallucinate non-existent web domains for legitimate brands. Threat actors register these AI-generated domains and host phishing pages or malware, targeting traffic mistakenly directed by AI tools. This novel attack vector, identified by Palo Alto Networks’ Unit 42, creates a difficult-to-detect supply chain threat by leveraging AI’s generative capabilities.