A documentary made entirely by AI just told me not to trust it. I watched a 27-minute film last night. The host, "Imogen Ashbery," walked me through superintelligence, the jobs ladder, data centres, AlphaFold, and deepfakes. Sharp, funny, well sourced. Then she told me she wasn't real. Never had a cup of tea. Every smirk statistically likely. Here is what I took from it as someone who runs Responsible AI for a living: 1. Disclosure works when it is part of the story, not a footnote. The reveal was the most persuasive moment in the film. 2. The ladder problem is real. Stanford's payroll data shows 22 to 25 year olds in AI-exposed jobs falling behind their peers. The bottom rung is not snapping. It is disappearing. We need to rebuild how juniors learn, not pretend the rung is still there. 3. Her three house rules are the whole of my transparency policy in one breath: if it is fake, say so. If it is built from a real person, pay them. If your boss calls asking for $25M, ring them back. I publish my own rules at rai-policy.arnao.ai. They are longer. Hers are better. What would make you trust a synthetic presenter? Sources in the first comment.