Zscaler says ransomware crews use generative AI for custom malware
Original title: Hackers Use Generative AI to Build Custom Malware and Bypass Traditional Security Detection
An October 16 analysis describes faster script changes, stronger phishing and quicker data discovery, with no named family or indicators.
Zscaler principal researcher Phil Hagen, writing on October 16, 2026, says attackers use generative AI to rewrite malware, craft phishing and classify data for theft or encryption. The analysis covers methods rather than a named malware family or a single confirmed campaign. Zscaler says signature matching weakens when code changes quickly and that smaller ransomware groups can now build custom tools. It describes AI-assisted development, not proof that payloads rewrite themselves while running.
Unconfirmed: The report describes methods and does not identify a named family or a confirmed campaign.
- No malware family, hashes, domains, URLs or filenames are identified.
- AI-written phishing can remove spelling and grammar mistakes used as warnings.
- Messages may persuade users to run remote monitoring and management software.
- Faster data classification can shorten the window before theft or encryption.
- Zscaler points to behavioral detection, combined data and faster investigation.
Impact: Custom malware, phishing and faster theft or encryption
- 1.Detect by program behavior rather than known-file signatures alone.
- Organizations
- Zscaler
From the original article
3 min read at cybersecuritynews.com
Hackers are using generative AI to build custom malware, rewrite harmful scripts, and make phishing messages harder to spot. The shift puts pressure on security tools that mainly depend on known file signatures, while helping ransomware operators move faster from their first foothold to data theft and encryption. The danger is not that every AI-generated […]
By Tushar Subhra Dutta · cybersecuritynews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.