Facebook Internally Reported Failure of Ranking Algorithm, Exposing Harmful Content to Viewers over Months

Laws on record for this risk area

Policy angle: Classified under Misinformation (False or misleading information) in the MIT AI Risk Repository taxonomy; 5 instruments on record cover this risk area.

• India DPDP Act India · Partially applicable · binding
• Law on Artificial Intelligence (2025) Kazakhstan · Adopted · binding
• Law No. 132/2025 on artificial intelligence Italy · In force · binding
• EU AI Act European Union · Partially applicable · binding
• Texas Responsible AI Governance Act (TRAIGA) Texas (United States) · In force · binding

Matched from the record's risk category and country to the instruments in effect recorded here: a law listed covers the area, which is not a finding that it applies to this incident. A reviewer can correct the match in the repository (data/external/incident_overrides.yaml).

News reports (4)

Titles link to the original publisher; report text is not reproduced here.

1. 31 Mar 2022
   A Facebook bug led to increased views of harmful content over six months

   theverge.com · Alex Heath
2. 31 Mar 2022
   Facebook boosted harmful posts due to 'massive ranking failure' bug

   protocol.com · Issie Lapowsky
3. 1 Apr 2022
   Meta admits Facebook bug led to a 'surge of misinformation'

   dailymail.co.uk · Jonathan Chadwick
4. 1 Apr 2022
   Facebook system designed to smother harmful misinformation actually spread it

   thedrum.com · John Glenday

Who was involved

Alleged deployer

Facebook

Alleged developer

Facebook

Alleged harmed party

Facebook Users

Classification (MIT AI Risk Repository taxonomy)

Risk domain

Misinformation

Risk subdomain

3.1 False or misleading information

Causal entity

AI

Intent

Unintentional

Timing

Post-deployment

Harm level

—

Sectors

—

Countries

—

Risk entries describing this failure mode

Entries from the MIT AI Risk Repository coded to subdomain 3.1.

• Pursuing Consistent Context

  "LLMs have been demonstrated to pursue consistent context [129]–[132], which may lead to erroneous generation when the prefixes contain false information. Typical examples include sycophancy [129], [130], false demonstra...

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Defective Decoding Process

  In general, LLMs employ the Transformer architecture [32] and generate content in an autoregressive manner, where the prediction of the next token is conditioned on the previously generated token sequence. Such a scheme...

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Noisy Training Data

  "Another important source of hallucinations is the noise in training data, which introduces errors in the knowledge stored in model parameters [111]–[113]. Generally, the training data inherently harbors misinformation....

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Untruthful Content

  "The LLM-generated content could contain inaccurate information"

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Hallucinations

  "LLMs generate nonsensical, untruthful, and factual incorrect content"

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Faithfulness Errors

  "The LLM-generated content could contain inaccurate information" which is is not true to the source material or input used

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Factuality Errors

  "The LLM-generated content could contain inaccurate information" which is factually incorrect

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

• Knowledge Gaps

  "Since the training corpora of LLMs can not contain all possible world knowledge [114]–[119], and it is challenging for LLMs to grasp the long-tail knowledge within their training data [120], [121], LLMs inherently posse...

  Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

Incidents in the same risk subdomain

• 12 May 2026Nonfiction Book 'The Future of Truth' Reportedly Included AI-Generated and Misattributed Quotations
• 6 May 2026Claude Console Reportedly Generated Phantom Legal Quotations in Trump Layoffs Court Filing
• 21 Apr 2026Purportedly AI-Enhanced Images of Iranian Women Protesters Were Reportedly Spread With Unverified Execution Claims
• 10 Apr 2026South Africa Draft National AI Policy Reportedly Included Fictitious References Believed to Be AI Hallucinations
• 8 Apr 2026Purportedly AI-Generated Image Reportedly Misled Daejeon Authorities Searching for Escaped Wolf Neukgu
• 17 Mar 2026Gemini and Grok Reportedly Misidentified Authentic Minab School-Strike Graveyard Photo as Unrelated Disaster Imagery

Other incidents involving Facebook

• 1 Dec 2022Facebook's Automated Moderation Allowed Ads Threatening Election Workers to be Posted
• 8 Dec 2021Facebook AI-Supported Moderation for Ads Failed to Detect Violating Content
• 21 Nov 2021Facebook’s Hate Speech Detection Algorithms Allegedly Disproportionately Failed to Remove Racist Content towards Minority Groups
• 20 Jul 2021Facebook's Automated Moderation Flagged Gardening Group's Language Use by Mistake
• 1 Mar 2021Facebook's Automated Tools Failed to Adequately Remove Hate Speech, Violence, and Incitement
• 15 Jan 2021Facebook's Automated Moderation Mistakenly Flagged Landmark's Name as Offensive