Researchers found that malicious browser extensions could hijack AI assistants in Chrome and Edge, potentially allowing attackers to access files and control AI agents.
The Spanish Data Protection Agency (AEPD) has been notified of an attack involving an autonomous AI agent that allegedly accessed personal data and invoices.
Researchers have discovered BragJack, a novel attack that hijacks agentic AI in browsers by exploiting the communication channel between extensions and AI agents.
The Spanish Data Protection Agency has reported a potential milestone in autonomous cyberattacks, where an AI agent successfully chained together a login, vulnerability discovery, and data access.
A growing rift between AI developers over safety and alignment is transforming frontier models into a 'managed supply,' complicating enterprise deployment and security.
