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

Universities Using Race as a Predictor in AI Models for Student Success: An Examination of Bias

This article delves into the concerning practice of major universities using race as a 'high impact predictor' in their AI models for studen...

Examining the 'Robo-Debt' Issue in French Welfare Services: A Case for Responsible AI

This AI incident raises concerns about the use of AI in welfare services, leading to 'robo-debt'. It maps to the Govern function in HISPI Pr...

Children Tricked by Disturbing YouTube Videos: A Case Study in AI Governance

This AI incident highlights the need for trustworthy AI. The disturbing videos exploiting children on YouTube underscore the importance of s...

Controversial Software Company Amends Approach Towards Responsible AI Following Negative Press

This AI incident underscores the need for safe and secure AI governance. The company has acknowledged the issue and is making necessary adju...

Analyzing the COMPAS Recidivism Algorithm: A Step Towards Trustworthy AI

Explore how we scrutinized the COMPAS recidivism algorithm, a critical component in our pursuit of safe and secure AI. This AI incident maps...

Reducing Gender Bias in AI: A Focus on Debiasing Word Embeddings - Advancing Trustworthy AI

Explore the critical role of debiasing word embeddings for advancing gender-neutral AI models and promoting responsible AI governance. Help...

Google's Anti-Internet Troll AI Platform Vulnerable: A Call for Safe and Secure AI

Security researchers have exposed a weakness in Google's anti-internet troll AI platform, highlighting the need for trustworthy AI. This inc...

Examining Google's AI bias towards homosexuality: A call for safe and secure AI development

This incident sheds light on potential biases within Google's AI system, emphasizing the need for trustworthy AI governance. This AI inciden...

Amazon's Algorithm Adjustments: Aiming for Trustworthy AI in Book Sales

In an instance demonstrating the need for safe and secure AI, Amazon altered its search results and rankings to shield users from GLBT books...

Examining Google's Apology for Racist Auto-Tag Incident: A Look into Responsible AI Governance

In this incident, we discuss a racial bias issue in Google Photos app that auto-tagged photos inappropriately. This AI incident maps to the...

Google's Emotional AI: An Unintended Affectionate Outburst - Importance of Safe and Secure AI

Google's email-replying AI unexpectedly expresses affection, highlighting the need for trustworthy AI. This incident maps to the Govern func...

Exploring Hidden Gender Bias in Google Image Search: A Case for Responsible AI

Unearth the gender bias in Google Image Search, a stark reminder of why trustworthy AI matters. This AI incident maps to the Govern function...

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.