How the TIME AI Audio Brief Was Built

At TIME, we’ve long embraced innovation in how our trusted journalism is delivered to readers, viewers, and audiences around the world. Today, we continue that tradition with the launch of our AI-generated audio briefing, a new format designed to make our reporting more accessible. This tool, created with technology partner ScaleAI, reflects TIME’s ongoing mission to meet people where they are—and where they’re going. 

The new audio briefing tool uses generative AI to present the day’s top news stories, as written by TIME’s reporters, into an engaging discussion between two AI bots, Henry and Lucy, named after one of the founders of TIME, Henry Luce. Our goal is to create an experience that makes our journalism more digestible and interactive. The audio briefing is designed to provide trusted and accurate reporting in a conversational tone. It is limited to discuss only articles previously published by TIME.

This tool is an additional way to distribute TIME’s journalism, which today can be found on our site, TIME.com, across social media, in print and on film and television platforms and channels, created by TIME Studios. This audio briefing represents another step in how we’re using AI to deliver the news. This is the second major launch for TIME AI, and we plan to continue to update and improve our AI products in the future. It’s our hope that all of this work increases accessibility and engagement with TIME’s journalism.

Here’s a closer look at how it works. 

How does the audio briefing work? 

This tool takes a selection of news stories from TIME’s flagship newsletter, The Brief, which is curated by a TIME editor. It uses GPT/o models, which are a family of large language models developed by OpenAI, a TIME strategic partner, that can generate content in multiple formats aside from just text. The GPT/o models from OpenAI were use for script generation and GPT4o-mini-TTS was used to convert the script into an audio file. This is a fully automated process using stories only published on TIME.com.

What voices are used in the audio? 

Henry and Lucy have distinct voices that were developed in collaboration with ScaleAI using OpenAI’s library of voices. They were tuned to reflect TIME’s overall editorial sensibility and to be responsive to the original reporting in the work that they cite. We’ve ensured that the AI hosts properly attribute both the work of TIME reporters as well as the sources featured within the reporting which are highlighted in The Brief.

What safeguards are in place? 

The models are instructed to ground discussions in the articles and to always attribute material to their origin in the articles. They are instructed not to deviate from the article content. There is, as we note in the audio player itself, a possibility of error.

How else does TIME use AI? 

In December 2024, we unveiled an AI toolbar specifically designed to answer questions about TIME’s Person of the Year based on TIME.com articles, other trusted sources and the tool’s general built-in knowledge. This product was engineered to provide reliable and accurate information while maintaining a safe and professional conversation. You can learn more about that the TIME AI toolbar here.