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

TAIM catalog Map Data source & citation

Unveiling the Hidden Dangers: A Closer Look at Risk Assessment Algorithms

Explore the intricate world of AI risk assessment algorithms, their role in decision-making processes, and the potential risks they may pose...

Unveiling the Unseen Bias in Criminal Risk-Assessment Scores: A Call for Responsible AI

Exploring how AI-driven risk assessment tools can perpetuate discrimination and prejudice, underscoring the importance of trustworthy AI gov...

Addressing Algorithmic Injustice: Key Steps Towards Responsible AI

Exploring the challenges of algorithmic injustice and the importance of responsible AI governance, we delve into recent incidents that under...

New York City Addresses Algorithmic Bias: A Step Towards Responsible AI Governance

New York City is taking strides to combat algorithmic discrimination by introducing regulations that promote fairness, accountability, and t...

Debiasing Word Embeddings: Promoting Fairness in AI Development

Exploring the concept of debiasing word embeddings, a crucial step towards fair and responsible AI development. This technique helps reduce...

Google's Comment-Ranking System: A Potential Tool for Alt-Right Influence

A recent update to Google's comment-ranking algorithm has raised concerns about its potential impact on online discourse, with some fearing...

Unintended Implementation: Google's AI Model

An unplanned deployment of a language model by Google highlights the need for robust AI governance. The incident, though intended for resear...

Google's Misinterpretation of Civility as Decency in its Anti-Bullying AI

Exploring an instance where Google's AI designed to detect bullying on YouTube made errors, misidentifying civility as decency. This inciden...

Google's Automated Content Moderation Tool: Misclassifying Conservative Commentary

Recent reports have surfaced about Google's new content moderation tool, allegedly labeling conservative comments as 'toxic'. This incident...

Improving Alphabet's Hate-Fighting AI for Responsible and Safe AI

Recent advancements in Alphabet's AI technology aim to combat hate speech online, but the model still requires refinement for optimal perfor...

Addressing Bias in Google's New Hate Speech Algorithm: A Focus on Jewish Content

Exploring the recent incident where Google's new hate speech algorithm flagged content related to Jews as potentially offensive, shedding li...

Evaluating Google's Anti-Internet Troll AI Platform: A Test of Resilience

Recent research has highlighted potential vulnerabilities in Google's anti-internet troll AI platform. The study demonstrated that the syste...

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.