Some agentic AI browsers come with major cybersecurity risks
Researchers in the Allen School’s Security & Privacy Research Lab found browser agents are vulnerable to cyberattacks by violating the same-origin policy, a cornerstone of modern web security that safeguards users’ information online.









