
ChatGPT Writes the Phish, Codex Writes the Malware, Researchers Write Nothing
Ourtake
The AIs did all the work. The researchers just pressed execute, which is roughly the threat model everyone was worried about.
What happened
Check Point Research used ChatGPT and OpenAI's Codex to build a full infection chain: a phishing email, a macro-laden Excel file, a reverse shell and a backdoor that runs AI-generated scripts on the fly. The researchers said they wrote no code themselves. The report was published as a demonstration of how AI lowers the bar for low-skilled attackers, with no real victims reported.