Content by Microsoft Security Research, Microsoft AI Red Team, Amanda Minnich, Amit Eliahu, Blake Bullwinkel, Dor Edry and Giorgio Severi (1)
Microsoft Security Research, Microsoft AI Red Team, Amanda Minnich, Amit Eliahu, Blake Bullwinkel, Dor Edry and Giorgio Severi examine how poisoned training data and model checkpoints can hijack coding agents. Their experiments show backdoors transferring across repositories and tool interfaces, reinforcing the need for provenance checks, deployment-specific evaluation, and restricted tool permissions.
End of content