Incident 281: YouTube's Algorithms Failed to Remove Violating Content Related to Suicide and Self-Harm

Description: Terms-of-service-violating videos related to suicide and self-harm reportedly bypassed YouTube’s content moderation algorithms, allegedly resulting in exposure of graphic content to young users via recommended videos.

Tools

New ReportNew ReportNew ResponseNew ResponseDiscoverDiscoverView HistoryView History

Entities

View all entities
Alleged: YouTube developed and deployed an AI system, which harmed YouTube young users and YouTube users.

Incident Stats

Incident ID
281
Report Count
3
Incident Date
2019-02-04
Editors
Khoa Lam
Applied Taxonomies
CSETv1, GMF, MIT

CSETv1 Taxonomy Classifications

Taxonomy Details

Incident Number

281

MIT Taxonomy Classifications

Machine-Classified
Taxonomy Details

Risk Subdomain

1.2. Exposure to toxic content

Risk Domain

  1. Discrimination and Toxicity

Entity

AI

Timing

Post-deployment

Intent

Unintentional

YouTube recommended self-harm videos to children as young as 13
telegraph.co.uk · 2019

Speaking to The Telegraph, a former Tumblr blogger, who asked for anonymity, said she had to stop her own depression and anxiety help blog after she found herself “falling down the rabbit hole of content that triggered negative emotions”.

“…

YouTube criticized for recommending 'self-harm' videos with graphic images
businessinsider.in · 2019

Potentially harmful content has reportedly slipped through the moderation algorithms again at YouTube.

According to a report by The Telegraph on Monday, YouTube has been recommending videos that contain graphic images of self-harm to users …

YouTube caught promoting deadly 'how to self harm' tutorials for youngsters aged 13
thesun.co.uk · 2019

YouTube has been caught recommending "dozens" of graphic videos relating to self-harm to children.

The hugely popular video sharing app has been blasted for promoting dangerous clips to users as young as 13.

YouTube is under fire for failin…

Variants

A "variant" is an incident that shares the same causative factors, produces similar harms, and involves the same intelligent systems as a known AI incident. Rather than index variants as entirely separate incidents, we list variations of incidents under the first similar incident submitted to the database. Unlike other submission types to the incident database, variants are not required to have reporting in evidence external to the Incident Database. Learn more from the research paper.

Similar Incidents

By textual similarity

Did our AI mess up? Flag the unrelated incidents

Google’s YouTube Kids App Presents Inappropriate Content

Google’s YouTube Kids App Presents Inappropriate Content

· 14 reports
Content Using Bestiality Thumbnails Allegedly Evaded YouTube’s Thumbnail Monitoring System

Content Using Bestiality Thumbnails Allegedly Evaded YouTube’s Thumbnail Monitoring System

· 2 reports
Amazon Censors Gay Books

Amazon Censors Gay Books

· 24 reports
Previous IncidentNext Incident