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

Analyzing the COMPAS Recidivism Algorithm: A Case Study for Safe and Trustworthy AI

Delve into our comprehensive analysis of the controversial COMPAS recidivism algorithm, shedding light on its implications for harm preventi...

Debiasing Word Embeddings: Breaking Gender Stereotypes in AI Models

Explore how AI models can reinforce gender stereotypes through word embeddings. Learn about techniques to debias these models, ensuring trus...

Study reveals bias and rigidity in AI's civility detection, emphasizing the need for safe and secure AI

A recent study highlights a concerning display of bias and inflexibility in an AI system designed for civility detection. This incident unde...

Exploring Google's AI Bias Incident Regarding Homosexuality - Promoting Responsible AI

Dive into this incident involving Google's AI, which demonstrated bias towards homosexuality. As we strive for trustworthy and unbiased AI,...

Amazon Modifies Search Algorithm to Ensure Trustworthy AI: Protecting Users from Inappropriate Content

This AI incident, which maps to the Govern function in HISPI Project Cerebellum's Trusted AI Model (TAIM), demonstrates Amazon's commitment...

Apology from Google over Racist Auto-Tag Incident Highlights Need for Responsible AI

Google has issued an apology following a controversial incident where its photo app inappropriately tagged images of African Americans with...

Google's Affectionate Email-Replying AI: Examining a Case Study in Safe and Secure AI

Investigate the instance where Google's AI responded to emails by expressing affection, revealing potential vulnerabilities in trustworthy A...

Unveiling the Unseen Gender Bias in Google Image Search: A Call for Responsible AI Governance

Exploring a significant AI incident, this article highlights the hidden gender bias in Google Image Search. It underscores the importance of...

Investigation into Critical Amazon Warehouse Bear Spray Incident: A Case for Safe and Secure AI in Workplace Environments

OSHA is currently investigating a bear spray accident at an Amazon warehouse that left a worker critical. This incident underscores the impo...

Exploring Potential Risks in Google's Ad-Targeting System: A Look at Responsible AI Practices

This AI incident raises questions about the ad-targeting system within Google, highlighting the need for trustworthy and safe AI practices....

Tesla on Autopilot Collides with Police Car: Driver Admittedly Watching a Movie - Underlining the Importance of Safe and Secure AI

This Tesla accident, where the driver was watching a movie instead of the road, underscores the critical role of responsible AI. This incide...

Exposing Limitations in AI: A Case Study on the Updated Turing Test

This AI incident highlights the progress and challenges in AI development, demonstrating the need for robust guardrails for artificial intel...

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.