1 threat
Most recently published first
Attackers can craft a malicious prompt that coerces the language model into executing arbitrary JavaScript in the context of the web page.
Page 1 of 1