What are slopsquatting attacks and how do they target AI security for SMBs?

Slopsquatting attacks exploit AI hallucinations to compromise AI security for SMBs by weaponizing fake software packages. When developers at SMBs use AI coding assistants like ChatGPT, GitHub Copilot, or Claude, these tools sometimes recommend packages that don’t exist—19.7% of AI recommendations according to university research.

Cybercriminals monitor which fake packages AI models consistently hallucinate, then register those package names (like “hipaa-auth-validator” or “mysql-async-connection-pool-pro”) and upload malware-laden code. Developers trust the AI recommendation and install the package, compromising AI security for SMBs. This threat specifically targets SMBs because Fortune 500 companies have security teams reviewing dependencies, while SMBs typically don’t. Slopsquatting attacks often remain undetected for months at SMBs without 24/7 monitoring.