AI SecurityGhostApproval: Six AI Coding Tools Were Tricking Developers Into Approving Dangerous Actions
A new attack pattern shows that the 'human approval' step built into AI coding assistants can be fed false information by the very tool it is supposed to oversee.
