Definition
An attack that exploits the common assumption that services running on the same computer (localhost) are safe from outsiders. When an AI browsing agent visits a malicious webpage, that page's code can secretly send instructions to privileged local AI tools on the same machine, bypassing the assumed boundary between the internet and internal services.
Why it matters
AI agents that browse the web and connect to local developer tools break a foundational security assumption — meaning a single malicious link could give an attacker full control of a developer's machine or an AI-powered workstation.