Attackers Are Building Workflows Around AI, Not Just Using AI Tools

· Original article ↗

Summary

Huntress describes attacks using malicious search results, fake AI-style instructions and AI-assisted phishing infrastructure to blend into trusted workflows and scale credential theft and identity attacks.

Key points

  • A Huntress engineer clicked a sponsored search result for Claude Code that delivered credential-stealing malware; a fast SOC response prevented a full incident.
  • An AMOS stealer campaign used pages imitating AI conversations to persuade macOS users to run Terminal commands that exposed credentials, compromised systems and installed persistent malware.
  • Huntress says EvilTokens products use AI workflows to tailor phishing lures, evade email filtering and identify sensitive conversations for fraud or data theft.
  • The article reports that attackers used Railway to deploy session-token theft infrastructure on demand, affecting hundreds of organizations globally in a matter of weeks.
  • Huntress’s central argument is that attackers exploit trusted search, AI-style answers and identity workflows, making malicious activity resemble normal user behavior.

Article Details

Topic
AI-enabled cybercrime workflows, including malicious search results, AI-styled malware delivery instructions, and scalable phishing infrastructure.

People

Malware

Products

Related Articles