Fable App's AI-Generated Reading Summaries Contained Biased and Offensive Commentary
Fable's AI-generated reading summaries reportedly contained biased and offensive commentary targeting readers based on race, gender, sexuality, and disability. After complaints, including from user Tiana Trammell who received a summary suggesting she read white authors, Fable removed all AI features from the app.
What happened
Fable, a book tracking app, generated personalized reading summaries using an AI model that produced offensive and biased language. On December 29, 2024, user Tiana Trammell received a summary that suggested she should read white authors, stating: 'Your journey dives deep into the heart of Black narratives and transformative tales, leaving mainstream stories gasping for air. Don't forget to surface for the occasional white author, okay?'
Multiple users reported receiving offensive summaries. One user who read books about people with disabilities was told her choices 'could earn an eye-roll from a sloth.' Another user received commentary suggesting his diverse reading made him wonder 'if you're ever in the mood for a straight, cis white man's perspective.'
Reported impact
Chris Gallello, head of product at Fable, acknowledged receiving complaints about 'very bigoted racist language' and called some results 'disturbing' given the app was intended to be a 'safe space' for readers. Fable's filters for offensive language and topics failed to prevent the offensive content.
Index analysis
The AI model was designed to create brief, personalized summaries drawn from book descriptions. The offensive summaries suggest the model generated biased assessments of readers' choices, with particular bias against readers of Black narratives and diverse literature.
Governance controls
Fable announced it would remove all AI-generated features from the app, including personalized summaries and year-end reading wraps. The company also introduced safeguards including disclosures that summaries were AI-generated, opt-out capability, and a thumbs-down button to flag problems.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.