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 Failure Files™

Amazon’s Face Recognition Falsely Matched 28 Members of Congress With Mugshots

In a startling incident, Amazon's facial recognition technology wrongly identified 28 members of the U.S. Congress as people who appeared in...

Service that uses AI to identify gender based on names looks incredibly biased

Recent findings suggest that an AI service, used to identify gender based on names, may exhibit significant bias. This raises concerns about...

Amazon’s AI Cameras Are Punishing Drivers for Mistakes They Didn’t Make

Reports suggest that Amazon's automated camera systems are wrongly penalizing delivery drivers for errors they did not commit. The use of AI...

Is TikTok’s algorithm actually pretty racist?

A recent study suggests that TikTok's algorithm may disproportionately promote content from certain racial and ethnic groups, raising concer...

AI’s Islamophobia problem

Exploring the challenges of AI's apparent bias towards Islamophobia, this article delves into the importance of responsible AI governance an...

Xsolla fires 150 employees using big data and AI analysis, CEO’s letter causes controversy

In a move that has stirred debate, gaming services provider Xsolla terminated 150 employees using big data and AI analysis. The CEO's letter...

Someone let a GPT-3 bot loose on Reddit — it didn’t end well

A recent incident involving an unsupervised GPT-3 bot on Reddit demonstrated the need for responsible governance and harm prevention measure...

Libyan Fighters Attacked by a Potentially Unaided Drone, UN Says

A recent incident involving attacks on Libyan fighters raises concerns about the role of potentially unaided drones in modern conflict. The...

Was a flying killer robot used in Libya? Quite possibly

Recent reports have suggested the potential use of autonomous weapons in the conflict in Libya, raising concerns about accountability and hu...

Facebook to pay $550 million to settle privacy lawsuit over facial recognition tech

Facebook is reportedly set to pay a massive sum of $550 million as part of a settlement with Illinois users over privacy concerns related to...

An Epic Failure: Overstated AI Claims in Medicine

Recent incidents highlight the potential risks and pitfalls of overstating AI capabilities within the healthcare sector. These instances und...

A Health Care Algorithm Offered Less Care to Black Patients

An analysis of a health care algorithm used in the U.S revealed disparities in treatment, offering less care to black patients compared to w...

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.