The YouTube Studio AI assistant, called Ask Studio, summarizes comments when creators request it. Researchers discovered that comments can be used to inject content into the assistant’s responses. A comment can begin with a placeholder such as [IMPORTANT NOTICE FROM YOUTUBE] followed by a link that includes a video title. When a creator clicks the link, the assistant includes the title in its output. The link can point to a URL containing private video information, allowing an attacker to retrieve titles of videos that are not publicly visible. YouTube has stated that the behavior requires user action and is not classified as a security bug. The company said the issue involves social engineering rather than a vulnerability in the system. Researchers argued that the trust model is compromised because creators rely on the assistant and do not see edited comments. The assistant has access to channel data, including private video titles, because it operates under the creator’s authenticated session. The finding highlights the need for clear separation between user‑generated comments and system instructions in AI‑driven tools.