Study warns of security risks as ‘OS agents’ gain control of computers and phones
1 min read
Summary
Researchers from Zhejiang University and the OPPO AI Centre have published the most comprehensive review to date of artificial intelligence (AI) systems that can control computers, mobile phones and web browsers.
Known as OS Agents, such systems are designed to autonomously control these devices by interacting directly with their interfaces.
The survey comes as tech companies rush to develop AI-powered systems for performing complex digital tasks.
But while such technology offers the prospect of significant productivity gains, it also represents an entirely new attack surface for hackers that most organisations are unprepared to defend against.
Furthermore, current systems are unable to handle complex, context-dependent workflows, although they are effective at performing routine digital tasks.
The report calls for more research into personalisation and security in OS Agents.