Unseeable prompt injections in screenshots: more vulnerabilities in Comet and other AI browsers

The full story

This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.

Vulnerabilities in AI Browsers Expose Users to Prompt Injection Attacks New vulnerabilities have been identified in AI browsers, particularly OpenAI's Atlas and Comet, which allow for prompt injection attacks. These attacks can manipulate AI commands through unintended text input, posing risks to user actions like shopping and web browsing. 2 sources 20