The Story The Epstein Files, AI journalism and the future of truth - The Sunday Story
22 snips
Mar 29, 2026 
James Ball, a tech journalist and PhD researcher in AI, and Adam Levy, creator of the AI-generated investigation The Epstein Files and builder of the Distill app. They discuss building an AI-native investigation, how documents were ingested and curated, technical safeguards and quality controls, audience reaction, legal risks and fact-checking, and whether AI can replace human judgement in investigations.
AI Snips
Chapters
Transcript
Episode notes
Creator Used Personal App To Launch Epstein Series
- Adam Levy built a personal app Distill to turn large document dumps into audio summaries and then applied it to the Epstein documents.
- He published The Epstein Files, an AI-native forensic audit, which hit 100,000 downloads in seven days and 2 million in 30 days.
AI Fits Where Human Reading Hits A Wall
- Large unstructured document dumps (3.5 million files) are impossible for humans to read in reasonable time, creating opportunity for AI to synthesize them.
- Adam treated the DOJ release as data to ingest, contextualize and surface focused storylines via AI rather than traditional manual reading.
Drive AI By Clear Questions Not Open Prompts
- Start AI investigations by curating explicit questions to 'interview' the data, not by letting models roam freely.
- Adam asked specific questions (Who was Epstein? What was the operation? What role did the island play?) and used answers to structure episodes.