Artificial Intelligence Policy Tracker
Search
Menu

AI incident ·

Microsoft AI Poll Allegedly Causes Reputational Harm of The Guardian Newspaper

17 news reports Snapshot 7 Sep 2026
On this page 9 sections

In brief

An AI system built and deployed by Microsoft allegedly harmed The Guardian and Family Of Lilie James.

Risk domain
Misinformation False or misleading information
Occurred
Coverage
17 reportsOct 2023 - Nov 2023

What happened

An AI-generated poll by Microsoft, displayed alongside a Guardian article, inappropriately speculated on the cause of Lilie James's death, leading to public backlash and alleged reputational damage for The Guardian. Microsoft acknowledged the issue, subsequently deactivating such polls and revising its AI content policies.

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.

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 (17)

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

  1. Microsoft accused of damaging Guardian’s reputation with AI-generated poll
    theguardian.com · Dan Milmo
  2. Inserted AI-generated Microsoft poll about woman’s death rankles The Guardian – Ars Technica
    arstechnica.com · Benj Edwards
  3. Microsoft AI inserted a distasteful poll into a news report about a woman’s death
    theverge.com · Wes Davis
  4. The Guardian criticizes Microsoft over insensitive AI-generated poll
    readwrite.com · Maxwell William
  5. The Guardian battles controversy over offensive AI-backed poll, blames Microsoft for it
    neowin.net · Anushe Fawaz
  6. Guardian accuses Microsoft of inserting ‘crass’ AI poll in story about woman’s death
    telegraph.co.uk · James Warrington
  7. AI-Generated Poll Speculating on Woman’s Death Sparks Fight Between Microsoft and Guardian Newspaper
    themessenger.com · Abubakar Idris
  8. Microsoft's AI News Poll Ticks Off UK Newspaper
    fool.com · Isobel Asher Hamilton
  9. Microsoft launches probe after AI poll on woman's death sparks blowback
    axios.com · Sara Fischer
  10. Microsoft under fire over AI-generated poll about death of 21-year-old woman on its news platform
    businessinsider.com · Polly Thompson, Haley Tenore
  11. Microsoft's AI Blunder: Inappropriate Poll Sparks Public Outrage
    boingboing.net · Mark Frauenfelder
  12. It's another AI failure for Microsoft News as it asks users to vote on how a young woman died
    windowscentral.com · Kevin Okemwa
  13. Microsoft Criticized for Embedding ‘Crass’ A.I. Poll Beside News Article
    nytimes.com · Jenny Gross
  14. Microsoft Runs Ghoulish AI Poll Next to Article About a Dead Woman
    futurism.com · Noor Al-Sibai
  15. Microsoft AI inserts ‘distressing’ poll in article
    ia.acs.org.au · Denham Sadler
  16. Microsoft Nixes AI-Generated News Polls After One Asked About Woman's Death
    extremetech.com · Josh Norem
  17. The Guardian accuses Microsoft of damaging its reputation with offensive AI-generated poll
    searchengineland.com · Nicola Agius

Who was involved

Alleged deployer
Microsoft
Alleged developer
Microsoft
Alleged harmed party
The Guardian, Family Of Lilie James

Classification (MIT AI Risk Repository taxonomy)

Risk domain
Misinformation
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

All incidents in this subdomain

Other incidents involving Microsoft

Source record: incident #612 on the AI Incident Database · all 17 reports