Leaking YouTube Creators Private Videos
Summary
The article exposes a prompt-injection vulnerability in YouTube Studio's AI assistant 'Ask Studio', showing how attacker comments can influence AI responses and potentially leak private video data. It argues that content should be treated as untrusted input and calls for stricter separation between user-generated prompts and system directives, highlighting privacy risks for creators. The author escalates the issue after YouTube declined to classify it as a bug, emphasizing the need for improved trust boundaries in AI-integrated features.