A Realistic Look at Implications of ChatGPT for Cybercrime

Marcus Hutchins, a well-known malware researcher, critically examines popular claims that ChatGPT will revolutionize cybercrime. He argues that ChatGPT is too limited for generating functional malware due to poor complex code generation, token limits, and inconsistent outputs. Claims about polymorphic malware bypassing antivirus are debunked since modern AV uses behavioral detection, not code signatures. Phishing email generation via ChatGPT is dismissed as trivial since phishing was never technically difficult and better resources already exist. Forum posts citing ChatGPT use in cybercrime are characterized as circular reporting driven by industry hype. The one realistic threat identified is LLMs potentially optimizing large-scale social engineering operations like troll farms or tech support scams.