Attackers can craft a malicious prompt that coerces the language model into executing arbitrary JavaScript in the context of the web page.
{ "nvd_published_at": "2024-08-07T23:15:41Z", "cwe_ids": [ "CWE-79" ], "severity": "MODERATE", "github_reviewed": true, "github_reviewed_at": "2024-08-08T16:30:17Z" }