Blogs
GhostGPT offers AI coding, phishing assistance for cybercriminals
Summary: In January 2025, Abnormal Security reported that GhostGPT was being marketed to criminals as an AI assistant for phishing and malicious coding.
The researchers tested the service and obtained a document-signing impersonation email. They described Telegram-based access and advertising that claimed conversations were not logged. Those privacy claims were promotional assertions, not independently demonstrated guarantees.
Abnormal suggested the service might use a jailbroken model or an open-source language model; its underlying technology was not established. A convincing email output does not, by itself, verify every advertised malware capability.
CNB Telecom perspective
Polished writing is not evidence that a message is genuine. Organizations should independently confirm unusual payment or document requests and combine identity checks, reporting procedures and technical email controls. This is CNB Telecom’s defensive interpretation, not an endorsement of the service or a claim that one security product is sufficient.
Original source
Independently summarized and reviewed by CNB Telecom
Historical report; reviewed October 11, 2026.