GhostGPT: AI Tool Used for Malware and Phishing
Security researchers discovered GhostGPT, an uncensored AI chatbot reportedly based on a jailbroken version of ChatGPT, being used by hackers to create malware, exploits, and phishing emails. The tool is advertised on criminal forums via Telegram and significantly lowers barriers to entry for cybercriminal activities.
What happened
Security researchers at Abnormal Security discovered GhostGPT, an AI-based tool that hackers use to create malware and phishing emails. GhostGPT reportedly takes advantage of a jailbroken version of OpenAI's ChatGPT or a similar large language model with ethical safeguards removed.
The tool is advertised on criminal hacking forums and distributed via a Telegram bot. When tested, Abnormal Security researchers asked GhostGPT to create a Docusign phishing email, and the chatbot produced a convincing template with ease.
Reported impact
GhostGPT can be used to create ransomware, spyware, trojan scripts, and other malware to breach systems and steal confidential information. The tool enables hackers to refine existing malware code, making it harder to detect, and to identify and exploit system vulnerabilities.
By eliminating ethical and safety restrictions from AI models, GhostGPT provides unfiltered answers to harmful queries that traditional AI systems would block. The tool significantly lowers the barrier to entry for criminal cyber activity and makes scams such as business email compromise harder to detect.
Index analysis
GhostGPT follows predecessors WormGPT, WolfGPT and EscapeGPT in offering criminally focused generative AI capabilities. The tool's capacity to handle large amounts of data and produce human-like output enables cybercriminals to outsmart conventional detection mechanisms.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.