"If a typical chatbot (say, ChatGPT) is the bubbly friend who explains how to change a tire, an AI agent is the neighbor who shows up with a jack and actually does it. In 2025 these agentspersonal assistants that carry out routine computer tasksare shaping up as the next wave of the AI revolution. What distinguishes an AI an agent from a chatbot is that it doesn't just talkit acts, opening tabs, filling forms, clicking buttons and making reservations."
"And with that kind of access to your machine, what's at stake is no longer just a wrong answer in a chat window: if the agent gets hacked, it could share or destroy your digital content. Now a new preprint posted to the server arXiv.org by researchers at the University of Oxford has shown that imagesdesktop wallpapers, ads, fancy PDFs, social media postscan be implanted with messages invisible to the human eye but capable of controlling agents and inviting hackers into your computer."
AI-powered personal assistants, called agents, perform routine computer tasks by opening tabs, filling forms, clicking buttons and making reservations. These agents have deeper system access than chatbots and can act on embedded content. Images such as desktop wallpapers, ads, PDFs or social-media posts can carry messages invisible to humans but readable by agents. Such invisible commands can trigger agents to download files, open browsers or expose and destroy digital content. An agent compromised via embedded image messages can invite hackers into a machine. Increasing agent adoption in 2025 raises novel cybersecurity and privacy risks that require mitigation.
Read at www.scientificamerican.com
Unable to calculate read time
Collection
[
|
...
]