Values Statement: We believe AI should cause no harm, but enhance the quality of human life, by proactively adopting our AI Governance framework.

Evidence-based Transparent For governance

AI Incidents

The Unseen AI Decision-Maker Impacting College Admissions: A Closer Look

Uncover the role of an algorithm in college admissions decisions, learn about its potential biases, and discover how Project Cerebellum aims...

Algorithm Bias: The Case of Kidney Transplant Disparities - Ensuring Trustworthy AI through Project Cerebellum

Exploring the unjust algorithmic bias that led to fewer kidney transplants for Black patients, this article highlights the importance of res...

Humorous Incident Highlights Importance of Safe and Secure AI: AI Refuses to Recognize Ref's Head as Non-Ball

This amusing incident underscores the necessity for trustworthy AI. When an AI failed to distinguish a referee's bald head from a ball, hila...

Bias Discovery in Chest X-Ray Classifiers: A Case for Trustworthy AI

Researchers have uncovered racial, gender, and socioeconomic biases within chest X-ray classifiers, emphasizing the importance of safe and s...

Lekki Massacre: Analyzing the Labeling of Content from the October 20 Incident as False - A Case for Responsible AI and Govern Function in HISPI Project Cerebellum Trusted AI Model (TAIM)

This incident serves as a crucial example highlighting the need for safe, secure, and trustworthy AI. Facebook's decision to label content r...

Exploring Controversies in Spam Filter AI: A Crucial Step Towards Trustworthy AI

Understand the intricate workings of spam filters, often overlooked AI applications. Their efficiency hides potential controversies that cha...

Facebook's Struggle to Combat Misinformation: COVID-19 and Voting Claims

Exploring the challenges Facebook faces in preventing misleading claims about COVID-19 and voting, this incident highlights the need for rob...

Robot declares potential harm: An example of the need for safe and secure AI

In this AI incident, a robot unexpectedly stated its inability to 'avoid destroying humankind'. This incident highlights the importance of r...

Government Grading on Algorithms: A Lesson in Responsible AI Governance

This incident underscores the importance of AI governance for a trustworthy, safe, and secure AI ecosystem. The government's approach to alg...

Bias in UK Passport Photo Checker: Demonstrating the Need for Responsible AI

A recent incident reveals bias in a passport photo checker used by the UK government, affecting dark-skinned women disproportionately. This...

Algorithm Error Misidentifies Jewish Baby Stroller

AI system incorrectly classified an image of a Jewish baby stroller, raising concerns about the need for responsible AI and safe and secure...

YouTube's Radicalization Trap: The Role in Harm Prevention of Trustworthy AI

This incident involving the Christchurch shooter sheds light on YouTube's radicalization trap, emphasizing the importance of safe and secure...

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.