Post

It concerns me that we spent decades building security best practices, principles of least privilege, account access controls, multi-factor authentication, passkeys, and other processes... And then all these 'genius' AI companies ignore it all and give their agents rampant access to everything.

2/6/26, 2:27 am0 reposts3 likes

Tabin Ahmad

@panicswitched.com

2 June

Facebook is out here with their AI Agent literally being scammed into giving away control of highly public and visible accounts via rudimentary prompt injection attacks and completely bypassing all common verification requirements is willllld.