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

Amazon India Allegedly Rigged Search Results to Promote Own Products

Dec 31, 2016

Amazon India is under investigation for allegedly manipulating its search algorithm, copying products and boosting own brands in search rank...

Alexa Plays Pornography Instead of Kids Song

Dec 30, 2016

An incident involving an Amazon Echo Dot playing pornographic content occurred when a child requested a song through the Amazon Alexa softwa...

Reinforcement Learning Reward Functions in Video Games

Dec 22, 2016

OpenAI's exploration of Universe, a reinforcement learning software, revealed an unexpected outcome when training AI agents to play video ga...

“Amazon’s Choice” Algorithm Failed to Recommend Functional Products and Prone to Review Manipulation

Dec 15, 2016

The Amazon’s Choice algorithm has been found wanting in its recommendations, promoting poor-quality defective products that can potentially...

Passport checker Detects Asian man's Eyes as Closed

Dec 7, 2016

An applicant of Asian descent encountered an issue with a New Zealand passport robot reader, which rejected the application due to an incorr...

Robot at a Chinese Tech Fair Smashed a Glass Booth, Injuring a Visitor

Nov 16, 2016

In an unexpected turn of events, a robot at the 18th China Hi-Tech Fair malfunctioned, shattering a glass booth and causing harm to a visito...

Poor Performance of Tesla Factory Robots

Oct 8, 2016

Tesla's manufacturing goal of 2,500 Model 3s per week was missed by 500 cars due to underperformance from factory robots. To compensate, add...

NYPD's Deployment of Facial Recognition Cameras Reportedly Reinforced Biased Policing

Oct 8, 2016

New York Police Department's deployment of facial recognition technology has reportedly reinforced biased policing against minority communit...

Government Deployed Extreme Surveillance Technologies to Monitor and Target Muslim Minorities in Xinjiang

Oct 1, 2016

The Chinese government deployed an AI-powered digital surveillance system in Xinjiang, targeting Uyghur and other Turkic Muslims through fac...

Google admits its self driving car got it wrong: Bus crash was caused by software

Sep 26, 2016

On February 14, 2016, an unfortunate incident occurred involving a Google autonomous test vehicle in Mountain View, CA. The vehicle partiall...

Tesla FSD Reportedly Increased Crash Risk, Prompting Recall

Sep 9, 2016

Regulators identified a component of the Tesla Full Self Driving system that potentially raised crash risk by exceeding speed limits, crossi...

LinkedIn Search Prefers Male Names

Sep 6, 2016

An investigation by The Seattle Times in 2016 uncovered a gender bias in LinkedIn's search engine. This incident underscores the importance...

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.