Sensemaker Can you spot the AI video?
Jan 2, 2026
Patricia Clarke, a Technology reporter at The Observer, dives into the complexities of AI-generated videos. She discusses the alarming scale of AI content on platforms like TikTok, highlighting billions of views on problematic posts. Patricia also examines how companies are responding, from watermarking to regulating AI media globally, and the challenges of accurately labeling AI creations. The conversation unveils the public's mixed feelings about AI content, alongside industry pushback advocating for human-produced material.
AI Snips
Chapters
Transcript
Episode notes
AI Content Is Massively Widespread
- AI-generated videos are increasingly believable and include fake news reports and sexualised imagery that spread widely.
- Forensic AI found roughly 43,000 AI posts on TikTok that amassed over 4.5 billion views, yet under 2% were labelled as AI.
Platforms Adopt Diverse Controls
- Platforms are responding differently: Spotify removed millions of spammy tracks and Steam requires AI disclosure for games.
- OpenAI's Sora 2 adds watermarks while platforms update policies to manage AI content proliferation.
Regulators Mandate Transparency
- Regulators worldwide are pushing transparency: the EU AI Act and rules in China, India, and South Korea mandate visible AI disclosure.
- These rules vary from screen-covering labels to mandatory symbols for chatbots and face swaps.