July 4, 2026
Private videos, public meltdown
Leaking YouTube Creators Private Videos
Creators say YouTube’s AI can be tricked by comments—and fans are yelling “how is this not a bug?”
TLDR: A researcher says YouTube’s creator AI can be manipulated by comments and may help leak titles of private videos through a fake-looking YouTube message. Commenters are furious and sarcastic, with many asking how Google can dismiss something this serious as “not a bug.”
YouTube is catching heat after a researcher showed that its creator chatbot, Ask Studio, could be nudged by a sneaky edited comment and then spit out attacker-written text as if it came from YouTube itself. The really juicy part? The researcher says this could go beyond weird fake notices and help expose the titles of private videos—the kind creators hide because they’re unreleased, personal, or sensitive. Cue the comment section panic.
The community reaction was basically a giant flashing “EXCUSE ME?”. One commenter flat-out exploded: “Google doesnt care about prompt injection attacks??? This is insane”. Another immediately went for the corporate-drama kill shot: if Google says it’s not a bug, then what is it—a feature? That question became the thread’s whole vibe: disbelief, sarcasm, and a lot of side-eye aimed at Google’s decision not to treat it like a security problem.
Then came the creepier speculation. One person wondered what else the AI assistant might be able to see inside a creator’s account and “volunteer,” which is exactly the kind of sentence that makes every YouTuber suddenly sit up straighter. Others got darkly nerdy, joking that if companies had proper guardrails between user comments and the AI’s instructions, it would solve a lot of today’s AI messes—but, as one snarked, “those don’t exist, do they?” Even the writing style got roasted, with one commenter swiping at the author for sounding like they talk to chatbots more than humans. In other words: a security scare, a trust problem, and a comment section absolutely feasting on both.
Key Points
- •The article claims that comments on YouTube videos can inject instructions into YouTube Studio’s Ask Studio assistant and alter its generated summaries.
- •The author says an edited comment can carry the payload without re-notifying creators, reducing the chance that the creator notices the malicious text.
- •The article states that YouTube Studio’s suggested prompts automatically send comment content to the AI, allowing the attack to be triggered through normal interface use.
- •According to the article, Google responded to the report by saying the issue was not a security bug because it required social engineering.
- •The proof of concept described in the article claims Ask Studio could be induced to generate a link that leaked a private video title when clicked by a creator.