About

The Dangerous Press archive brings together digitized issues of African American newspapers from the early twentieth century, spanning the years 1900 to 1930. This period witnessed the Great Migration, World War I, the Red Summer of 1919, the Harlem Renaissance, and the rise of the “New Negro” movement—all documented in real time by the Black press.

These newspapers served as vital organs of information, advocacy, and community building for African Americans during an era of segregation, disenfranchisement, and systemic violence. They reported on local and national news often ignored or misrepresented by white-owned media, championed civil rights causes, celebrated cultural achievements, and provided a platform for Black voices and perspectives.

The name Dangerous Press honors this legacy and the risks these publications faced. During World War I, federal agents labeled the Chicago Defender “the most dangerous of all Negro journals” because of its unflinching coverage of lynching, segregation, and northern migration. Its editor, Robert S. Abbott, and other Black journalists were monitored, threatened, and censored for their outspoken defense of racial justice. What made these newspapers “dangerous” was not disloyalty, but their insistence on truth in an age of repression—their power to awaken, mobilize, and connect Black readers across the nation.

This collection draws primarily from the Library of Congress’s Chronicling America digital newspaper archive, supplemented with materials from other sources. The archive is a work in progress, with new issues and newspapers being added regularly.

To make these newspapers searchable, each page image goes through an automated process that converts the printed text into digital text, a technology known as optical character recognition, or OCR. First, a layout detection model identifies where text, headlines, and images appear on the page. Then, a vision-language AI model called GLM-OCR reads each text region. This approach produces significantly more accurate results than older OCR methods, but it is not perfect, especially on pages with heavy ink bleed, faded print, or unusual layouts. Older OCR tools relied primarily on visual pattern matching, which often produced garbled text when the print was unclear. Newer models like GLM-OCR consider both the marks on the page and the surrounding words to predict what makes sense, which improves accuracy but also means they can occasionally hallucinate plausible-sounding text that does not actually appear on the page. The searchable text should be treated as a useful but imperfect representation of the original printed page. OCR processing is ongoing, and accuracy will continue to improve as the tools and methods are refined.

Collection Overview

The archive currently includes issues from 77 newspapers published between 1900 and 1935.

The Dangerous Press archive is an ongoing project under active development. New newspapers and issues are being added regularly. If you have questions, suggestions, or encounter any issues, please contact Neal Caren.