Infosecurity reports on Wiz’s GhostApproval research, which found that six AI coding assistants could be tricked through symlinks in malicious repositories into writing to sensitive local files after a user approved an apparently safe change.
GhostApproval Flaw Hits Six Major AI Coding Assistants
Infosecurity reports on Wiz's GhostApproval research, which found that six AI coding assistants could be tricked through symlinks in malicious repositories into writing to sensitive local files after a user approved an apparently safe change.
Source: Infosecurity Magazine