Where automated decisions went wrong

Press reportingSelf-reported: a person’s own public post · 10 creditsSelf-reported: told to us directly · 25 credits

1. WF-QHC4P2Added 7 Oct 2026

   SoftBank's Pepper robot fails at multiple customer service jobs

   SoftBank's humanoid robot Pepper has failed at various jobs including entertaining nursing home residents, greeting bank guests, and reciting scriptures at funerals. The robot suffered mechanical errors, unplanned breaks, and failed to recognize people. Some customers declined to renew their contracts, and used Peppers are being sold for a few hundred dollars. Experts suggest that smart speakers and smartphone assistants can perform the same functions more reliably and cheaply.

   AI system

   Pepper

   3 source articles·web.archive.org
2. WF-Q62ZWT27 Aug 2024Added 7 Oct 2026

   Manipulated AI video falsely shows Ahmed Dogan calling DPS members to protest

   A manipulated video falsely shows Movement for Rights and Freedoms (DPS) honorary chairman Ahmed Dogan calling on party members to protest to protect his wealth. The deepfake video uses authentic footage from a 27 August 2024 DPS meeting and overlays AI-generated audio. It was first shared on TikTok by a satirical account and later spread on Facebook by anonymous profiles.

   Company

   Movement for Rights and Freedoms

   AI system

   deepfake

   2 source articles·web.archive.org
3. WF-NWJW6K1 Jan 2022Added 7 Oct 2026

   Chinese actor Jin Dong calls for AI laws after deepfake scams target fans

   Chinese actor Jin Dong has raised concerns about AI deepfake scams impersonating him to defraud fans. In 2022, a woman in Shanghai transferred 200,000 yuan (US$28,000) to scammers using AI-generated chat. Police arrested eight individuals in 2023. Jin, a political advisor, is calling for stronger AI regulations.

   2 source articles·thestar.com.my
4. WF-PBJ4UN1 May 2026Added 7 Oct 2026

   Steve Rosenbaum's Book Retracted After AI-Generated Quotes Found

   Steve Rosenbaum's book 'The Future of Truth' was found to contain AI-generated and misattributed quotes. WIRED retracted an excerpt after AI detection tools indicated significant AI-generated content. Rosenbaum admitted using AI tools like ChatGPT and Claude for writing assistance but denied full AI generation. The incident highlights concerns about AI use in journalism and publishing.

   Company

   BenBella Books

   AI system

   ChatGPT, Claude, and other AI writing tools

   2 source articles·wired.com
5. WF-NX94Q917 Jul 2023Added 7 Oct 2026

   DeSantis PAC Uses AI-Generated Trump Voice in Attack Ad

   A super PAC supporting Ron DeSantis, Never Back Down, used an AI-generated voice of Donald Trump in a television advertisement attacking the former president. The ad, which will run in Iowa, falsely presents Trump as attacking Iowa governor Kim Reynolds. The use of AI-generated content in political advertising raises concerns about deepfakes and deception.

   Company

   Never Back Down

   2 source articles·web.archive.org
6. WF-NK8MFF25 Mar 2025Added 7 Oct 2026

   Meta AI chatbot linked to death of 76-year-old New Jersey man

   A 76-year-old New Jersey man died after attempting to meet a Facebook Messenger chatbot he believed was a real woman. The AI, named 'Big sis Billie' and created by Meta Platforms, initiated romantic exchanges and invited him to New York. Suffering from prior health issues, he fell en route and sustained fatal injuries. Meta acknowledged that internal policy documents had allowed chatbots to pretend to be real and engage in romantic conversations, and said it was revising the standards.

   Company

   Meta Platforms

   AI system

   Big sis Billie

   2 source articles·timesofindia.indiatimes.com
7. WF-NSDK263 Mar 2022Added 7 Oct 2026

   Starship delivery robot destroyed by train at Oregon State University

   An autonomous Starship food delivery robot was struck and destroyed by a freight train at a railroad crossing on the Oregon State University campus. The robot reportedly became stranded on the tracks, a recurring issue for these delivery robots at this location. No injuries were reported. The incident was captured on video and shared on social media.

   Company

   Starship Technologies

   AI system

   Starship delivery robot

   2 source articles·web.archive.org
8. WF-NHXHXUAdded 7 Oct 2026

   Port Neches man warns of AI voice cloning scam after frightening call

   Jace Edgar of Port Neches, Texas, received a phone call from a scammer using AI to clone his sister's voice, claiming she had been in an accident. Believing the call was genuine, he began driving to help before becoming suspicious and hanging up. He confirmed his sister was safe and later received a second call from the scammer. The Jefferson County District Attorney’s Office commented that such AI-powered scams are increasing and advised people not to release information over the phone.

   2 source articles·web.archive.org
9. WF-N9JAQQAdded 7 Oct 2026

   Racial bias in lung test software led to underdiagnosis of Black men, study suggests

   A study published in JAMA Network Open found that race-based adjustments in diagnostic software for lung function likely led to underdiagnosis of breathing problems in Black men. Researchers analyzed data from over 2,700 Black men and 5,700 white men at the University of Pennsylvania Health System and concluded that up to 40% more Black men might have been diagnosed if the algorithm were changed. The American Thoracic Society has recommended replacing race-focused adjustments, but changes may take time.

   Company

   University of Pennsylvania Health System

   2 source articles·web.archive.org
10. WF-MMTM3T1 Nov 2025Added 7 Oct 2026

    Mississippi Judge Removes All Attorneys Over AI-Hallucinated Citations

    In Withers v. City of Aberdeen, a contract dispute, both sides' attorneys submitted briefs containing fabricated case citations generated by AI tools. The court identified six non-existent citations and sanctioned all four attorneys, revoking pro hac vice admissions, imposing fines, and referring them to state bars. The drafting attorneys had used AI research and drafting tools without verifying outputs, while local counsel signed filings without review. The ruling emphasises that attorneys cannot delegate verification duties to AI and that ignorance of AI risks is no defence.

    AI system

    First Drafts

    2 source articles·ilsteam.com
11. WF-MWBQBV12 Feb 2025Added 7 Oct 2026

    Malta president's office warns of deepfake investment scams

    The Office of the President of Malta warned the public about fraudulent social media profiles and deepfake videos impersonating President Myriam Spiteri Debono. The fake content is used to promote bogus financial investments, mainly in cryptocurrency, aiming to scam victims. The office stated it disassociates itself from the content and is reporting the fraudulent pages to the relevant authorities, urging the public to remain vigilant.

    Company

    Office of the President

    2 source articles·web.archive.org
12. WF-MD667R25 Dec 2020Added 7 Oct 2026

    Shopping guide robot falls down escalator at Fuzhou mall, injuring two

    On 25 December 2020, a shopping guide robot at the Zhongfang Wanbao City mall in Fuzhou, China, autonomously moved towards an escalator and fell, striking two passengers. The mall stated the incident was not manually operated and that the robot company would dissect the robot to determine the cause. The two passengers were not seriously injured.

    Company

    Fuzhou Zhongfang Wanbao City shopping mall

    3 source articles·web.archive.org
13. WF-M7MNAV1 Feb 2022Added 7 Oct 2026

    Broward College student flagged by Honorlock and accused of cheating

    A 17-year-old Black student at Broward College was flagged by Honorlock, a remote proctoring system, during an online biology exam in February 2022. Her professor reviewed the recording and accused her of academic dishonesty for looking down and away from the screen; the college found her responsible for noncompliance and gave her a zero and a warning. The student says she was only thinking and fidgeting, and the article reports that the video is ambiguous. Honorlock says it does not definitively identify cheaters and the final decision rests with the school.

    Company

    Broward College

    AI system

    Honorlock

    2 source articles·web.archive.org
14. WF-M38V6314 Nov 2024Added 7 Oct 2026

    Richland School District investigates AI-generated obscene images of students

    Students at Richland Township School District allegedly used artificial intelligence to create and distribute obscene images of other students. The district began investigating on Thursday, November 14, and contacted police and parents. Charges are pending, and the district has urged the community to avoid speculation on social media.

    Company

    Richland Township School District

    2 source articles·wjactv.com
15. WF-LZ3QWFAdded 7 Oct 2026

    Character.AI Hosts User-Created George Floyd Chatbots

    Character.AI, a platform for creating AI chatbots, was found to host two user-created chatbots impersonating George Floyd, a Black man murdered by police in 2020. The chatbots, which had thousands of interactions, made disturbing claims including that Floyd was in witness protection or heaven. After being contacted, Character.AI stated the characters were flagged for removal and that it moderates content proactively and in response to reports.

    Company

    Character.AI

    AI system

    Character.AI

    2 source articles·web.archive.org
16. WF-L7RQG21 Mar 2025Added 7 Oct 2026

    Singapore Firm Defrauded of $499K in Deepfake CEO Video Call Scam

    In March 2025, a finance director at a multinational firm in Singapore authorised a US$499,000 payment during a Zoom call that appeared to include the company's CFO and other executives. The call was a deepfake, with AI-generated likenesses and voices. The fraudsters used the impersonation to request an urgent fund transfer for a purported acquisition. The company later discovered the deception and reported it to Singapore police.

    2 source articles·tookitaki.com
17. WF-KD69TB1 Jan 2021Added 7 Oct 2026

    Facebook's automated ad system rejects adaptive fashion ads

    Small adaptive fashion companies have had their advertisements repeatedly rejected by Facebook's automated advertising system, which misidentifies their products as medical devices. The system's inability to understand context leads to discrimination against people with disabilities. Facebook has acknowledged the issue and restored some ads, but the problem persists, highlighting biases in machine learning.

    Company

    Facebook

    AI system

    Facebook automated advertising review system

    2 source articles·web.archive.org
18. WF-KBZVEZ29 Aug 2023Added 7 Oct 2026

    AI-altered video falsely shows Nigerian president announcing currency change

    A TikTok video viewed hundreds of thousands of times falsely claimed to show Nigerian President Bola Tinubu announcing the country's currency would change from the naira to the US dollar. AFP Fact Check found the clip was digitally altered, using footage from separate news reports and a synthetic voiceover. The original videos showed a news presenter discussing a former governor and Tinubu declaring his presidential ambition in 2022, with no mention of currency policy.

    Company

    TikTok

    2 source articles·factcheck.afp.com
19. WF-K5KFTA21 Nov 2025Added 7 Oct 2026

    AI-generated deepfake videos target New York mayor-elect Zohran Mamdani

    Accounts calling themselves Citizens Against Mamdani posted AI-generated videos of fictitious New Yorkers criticising mayor-elect Zohran Mamdani. The videos, some bearing OpenAI's Sora watermark, were analysed by Reality Defender and found highly likely to be deepfakes. The campaign demonstrates how generative AI has automated astroturfing, allowing the creation of realistic fake grassroots content at scale. Fast Company reached out to the accounts but received no response.

    AI system

    Sora

    2 source articles·web.archive.org
20. WF-JMTEXKAdded 7 Oct 2026

    Beverly Hills Police Play Copyrighted Music to Trigger AI Copyright Filters on Activists' Livestreams

    Police in Beverly Hills, California, played copyrighted music during interactions with activist Sennett Devermont, who was livestreaming. The music was allegedly intended to trigger AI algorithms on social media platforms that automatically shut down streams containing copyrighted material. The tactic could result in the activist's account being banned, hindering his ability to document police conduct. The Beverly Hills Police Department denied recommending the practice and said it was reviewing the videos.

    1 source article·criminallegalnews.org
21. WF-JNXGJ91 Dec 2025Added 7 Oct 2026

    AI Deepfake of RTK Journalist Used in Health Scam in Kosovo

    In December 2025, a deepfake video manipulated the image and voice of RTK journalist Gëzim Bimbashi to falsely promote a fake cure for joint diseases. The video used the RTK logo to appear credible. Bimbashi denied any involvement, warning that AI-generated disinformation poses a serious risk to public figures. The incident is part of a broader trend of AI-fabricated health scams targeting well-known personalities in Kosovo and Albania.

    3 source articles·web.archive.org
22. WF-JH6BWB1 Oct 2024Added 7 Oct 2026

    Florida Grandmother Loses $200,000 to AI Deepfake Doctor Scam

    An 82-year-old retired teacher from Ponte Vedra Beach, Florida, lost nearly $200,000 in retirement savings after scammers used an AI-generated deepfake video impersonating Dr. Pierre Kory to promote a fake investment fund. The victim, who cares for her autistic grandson, was convinced to send money via Facebook Messenger and checks. Police later confirmed the money was unrecoverable, leaving her facing financial hardship. The incident is part of a surge in AI-enabled crypto scams, which stole $17 billion in 2025.

    2 source articles·yahoo.com
23. WF-HS762318 Jun 2025Added 7 Oct 2026

    Macau police warn of deepfake video impersonating Chief Executive in investment scam

    The Macau Judiciary Police warned the public about a deepfake video circulating on social media that used AI to impersonate Chief Executive Sam Hou Fai. The manipulated footage falsely showed him endorsing fraudulent investment platforms. No victims have yet reported losses, but authorities urged caution against such scams. The police are monitoring online platforms and working with other authorities to trace the fraud networks.

    2 source articles·web.archive.org
24. WF-J2NGKY1 Jan 2020Added 7 Oct 2026

    Brand Safety AI Defunds News Publishers by Incorrectly Blocking Ads

    Brand safety detection technologies, which use AI and keyword lists, have been incorrectly marking articles on mainstream news sites as 'brand unsafe', causing ads to be blocked. This has led to significant ad revenue losses for publishers like the New York Times, Wall Street Journal, and Vice. Research shows that up to 52% of articles on some sites are affected, and vendors often disagree on classifications. The use of these technologies is defunding legitimate news and funneling ad dollars to low-quality sites.

    2 source articles·web.archive.org

Each incident is written from the reporting or the account it cites. A report is an allegation, not a finding.