JupyterLab's ImageViewer component can be tricked into executing arbitrary JavaScript by loading a specially crafted SVG image. The flaw occurs when URL.createObjectURL is used and the blob URL is revoked too early, letting the image retain an executable same‑origin context. When the image is opened and then redirected to a new browser tab, the embedded script runs with the privileges of the JupyterLab server, allowing remote code execution. The vulnerability is a classic cross‑site scripting flaw classified as CWE‑79.

The issue affects JupyterLab releases prior to 4.5.10 and 4.6.2. Any installation using those versions is vulnerable when the image viewer is used to open malicious SVG files. The vulnerability was identified in the repository and has been fixed in newer releases.

The likely attack vector requires an attacker to supply a malicious SVG file that the victim opens in the image viewer and then opens in a new tab. The exploitation condition therefore depends on user interaction or compromised image files. The CVSS score of 7.5 indicates a high severity, and although no EPSS score is available, the lack of listing in CISA KEV suggests the issue has not yet been widely exploited. Nevertheless, the possibility of remote code execution warrants immediate mitigation.