CVE-2026-102832

A MODERATE severity vulnerability was found in JupyterLab, which is bundled in Red Hat OpenShift AI (RHOAI) workbench, notebook, and training container images. When a trust confirmation prompt is open, the markdownviewer:trust and notebook:trust command-link handlers resolve the document to trust from the currently active widget rather than the document whose command button was clicked. If the user switches to a different document while the prompt is open and then chooses Trust, JupyterLab marks the newly focused document as trusted while leaving the clicked document untrusted. An attacker could exploit this by convincing a user to open an untrusted notebook that unintentionally receives trusted status, allowing commands embedded in Markdown files or notebook outputs to run without further prompts; trusting a notebook also marks all of its cells trusted and persists the trust state. Exploitation requires no privileges but does require user interaction and a specific sequence of actions, which is reflected in the CVSS 3.1 score of 4.2 (AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:L/A:N); this matches the upstream severity assessment. Because the impact is MODERATE and the CVSS score is below 7.0, Red Hat has deferred the fix for the affected RHOAI container images, as trackers are not filed for MODERATE severity flaws without a higher CVSS score. The residual risk is limited by the high attack complexity and the required user interaction, but a successful attack could allow embedded commands in an attacker-supplied document to execute in the user's session with low confidentiality and integrity impact.