Sudipta Chattopadhyay

ASSET Research Group and Sudipta Chattopadhyay published a proof-of-concept attack called Ghostcommit that hides a prompt-injection instruction inside a PNG image, passes AI code review, and later ind…

Researchers at the University of Missouri-Kansas City developed 'Ghostcommit', an attack that hides prompt injection instructions inside PNG images to trick AI code reviewers into approving malicious …