AI-Generated Podcast ‘Epstein Files’ Ranks in Top 10 Charts
AI-Powered Podcast Dives Deep into Epstein Files
The Jeffrey Epstein case, already a fixture in headlines, is now the subject of an unusual investigative effort: a podcast called The Epstein Files that is entirely generated by artificial intelligence. The podcast has quickly gained traction, reaching the top 10 in the series rankings on Apple Podcasts and the top 30 on Spotify within days of its launch in early February .
What sets The Epstein Files apart is not just its subject matter, but its creator. Adam Levy, an entrepreneur with a background in data products and content creation, launched the podcast without prior journalism experience. He built an automated system to process the massive trove of documents released in connection with the case – over 3.5 million files – and transform them into podcast episodes.
From Data Dump to Daily Episodes
Levy’s approach involved building a pipeline that ingests the raw files, extracts text from emails and images, cross-references sources, and produces scripted episodes narrated by AI-generated voices. He described spending 48 hours, working 14- to 16-hour days, to initially “hack” together the automated system. The podcast currently releases two new episodes daily, a pace traditional newsrooms struggle to match.
The technical architecture relies on a combination of large language models (LLMs) from companies like Anthropic (Claude), Google, and OpenAI. These models are used to connect names, places, themes, and timelines within the document set. A “confidence score of veracity” is applied to connections before they are included in an episode, aiming to ensure factual accuracy. Levy also incorporates data from the Internet Archive and Google Pinpoint, a tool used by other investigators to index portions of the files, as well as projects like Jmail, which presents the Epstein Files emails in a user-friendly inbox format.
Referencing these sources was a deliberate strategy to address concerns about AI “hallucinations” – the tendency of LLMs to generate false or misleading information. “Everybody’s quite skeptical of AI,” Levy explained, “It was really important to reference all the sources that were used to basically construct the episode.”
A New Model for Investigative Journalism?
The podcast’s rapid rise in popularity has sparked discussion about the potential of AI in investigative journalism. Emily Bell, founding director of the Tow Center for Digital Journalism at Columbia University, noted that while the podcast is “obviously AI” to those familiar with the technology, it successfully “piques curiosity” and offers a compelling listening experience. She characterized the initial episodes as “listenable,” but also pointed to the potential for the format to become “tedious” over time.
The core appeal, according to Levy, is a direct and factual presentation of information. “People just want no bullshit,” he says. “Strip the emotion, strip the bullshit, strip everything away—just tell me things for what they are and when you tell it to me, help me understand the facts.”
Technical Underpinnings and Data Scale
The sheer scale of the data involved is a key factor driving the need for AI-assisted analysis. With over 3 million pages of documents – including DOJ files, court records, flight logs, and public resources – traditional journalistic methods are simply too slow and resource-intensive to process the information effectively. The AI architecture is designed to identify relevant passages, cross-reference names and dates, and synthesize findings into digestible episodes.
The podcast covers a wide range of topics related to the Epstein case, including his rise to wealth, his network of associates, the locations where crimes occurred, the controversial 2008 plea deal, his death in federal custody, the trial of Ghislaine Maxwell, and the remaining unanswered questions. The creators emphasize that the content is based on documented facts, processed at scale, and presented with journalistic rigor.
Podcast Charts and Industry Recognition
As of , The Epstein Files has published 97 episodes and garnered over 700,000 downloads. It currently ranks among the top subscriber shows on Apple Podcasts, alongside established programs like The Daily and Dateline NBC. The podcast also appears as a top series, ranking alongside Betrayal Season 5 and The Cult Queen of Canada from Uncover. Apple Podcasts’ charts for show The Joe Rogan Experience as the top show while Good Hang with Amy Poehler was the top new show.
The success of The Epstein Files highlights a growing trend in podcasting: the use of AI to automate content creation and analysis. While the long-term impact of this approach remains to be seen, it represents a significant shift in the media landscape and raises important questions about the future of investigative journalism.