Incidents gathered from public reporting around the world. Each one links to the articles it came from. None of it is a finding that anyone broke the law.
Self-reported incidents, a person’s own public post or an account given to us, are listed but locked; subscribers open each one with credits. Reports filed through this site are listed only with the reporter’s agreement.
Press reportingSelf-reported: a person’s own public post · 10 creditsSelf-reported: told to us directly · 25 credits
Labour MP Karl Turner shared a doctored image on X of Prime Minister Rishi Sunak holding a badly pulled pint. The image was a deepfake created using AI. Turner later apologised for sharing the image. The incident highlights the misuse of AI-generated content in politics.
Zoom modified its Terms of Service in March to permit training its AI models on user audio, video, and chat content. Users are not given a real opt-out; when the AI features are enabled, participants are told they can only leave the meeting or agree. The company responded with a blog post and statement clarifying that it will not use content without consent, but critics argue the choice is not meaningful. The backlash raises concerns about privacy and trust.
Jane Friedman, an author, says AI-generated books were uploaded to Amazon and Goodreads under her name without her consent. She reported the books to Amazon, which initially refused to remove them unless she provided a trademark registration number. After she publicised the issue, the fraudulent titles were removed from both platforms. She says the books misled readers and put her reputation at risk.
Brave, a privacy-focused web browser, is accused of selling copyrighted data from its Brave Search API to train artificial intelligence models without proper licensing. The allegations, reported by Stack Diary, claim that Brave enables third parties to ingest copyrighted content through its API. Brave's head of search defended the practice, stating that the rights issues relate to search engine results, not the content itself. The controversy raises questions about ethical data use and transparency in the tech industry.
Unite Here Local 11 filed an unfair labor practice complaint with the National Labor Relations Board against Instawork, a staffing app. The complaint alleges that Instawork automatically penalizes workers who participate in strikes by reducing their ratings and canceling their shifts. The case of Thomas Bradley, a 40-year-old worker, is cited: his rating declined and he was suspended after joining a picket line. Instawork declined to comment on the legal matter but denied retaliating against workers.
Researchers at Indiana University identified a botnet of 1,140 Twitter accounts, dubbed 'fox8', that appears to use ChatGPT to generate human-like content promoting crypto, blockchain, and NFT websites. The accounts were discovered because they accidentally posted the phrase 'as an ai language model' in tweets. The researchers warn that this botnet is likely the tip of the iceberg and that LLM-powered bots could be used for disinformation and political manipulation.
The journalist Andrey Kaganskikh found that access to Moscow's surveillance cameras and their facial recognition technology was being sold illegally. Law enforcement officials and government bureaucrats were said to sell live feeds and stored footage through the YTKD data-processing centre, with full access costing around 30,000 roubles. When he tested the face-recognition search with his own photograph, the system returned 238 matches from 140 different cameras at 67 percent accuracy, none of which was actually him. No convictions for such sales are recorded.
noyb filed a complaint against Ryanair with the Spanish Data Protection Authority (AEPD) on 27 July 2023. A customer who booked a Ryanair flight through online travel agency eDreams was required to undergo a facial recognition verification process or go to the check-in counter more than two hours before departure. The customer was charged a small fee for the verification. Ryanair outsources the facial recognition to GetID. noyb alleges that the process violates GDPR because consent is not valid and the purpose is to discourage bookings through third-party agents.
AI-generated images purporting to show Pope Francis wrapped in an LGBTQ pride flag were shared on social media. The images, created using Midjourney, contain visual inconsistencies indicating they are fabricated. AFP fact-checked the images and found no evidence they are genuine.
Rona Wang attempted to use an AI editing tool to generate a LinkedIn profile photo. Instead, the tool produced an unintended image. Wang shared the result on Twitter, expressing dissatisfaction.
Chris Robinson, then 75, applied for a senior living apartment in California. The property manager used an AI screening program from TransUnion, which assigned him a low score based on a mistaken conviction for littering that belonged to a different person with the same name in Texas. Robinson lost the apartment and his application fee. A federal class-action lawsuit against TransUnion moved toward a $11.5 million settlement in 2023.
The Maltese government has appointed a new board to run Safe City Malta, the Huawei-developed facial-recognition surveillance project, despite saying in January that the company would be wound down. The board, including Omar Debono, Joseph Cauchi and Nadine Mizzi, was appointed for another year until June 2024. The government had earlier dropped the facial-recognition element over concerns that it would breach Maltese and EU law, and the article reports that nothing has been delivered.
Cigna, one of the largest health insurers, used an algorithm called PXDX to automatically deny medical claims. The system flagged mismatches between diagnoses and procedures, and company doctors signed off on denials in batches without reviewing patient files. One patient, Nick van Terheyden, was denied coverage for a $350 vitamin D test that his doctor had ordered. Former employees said the system saved Cigna billions of dollars but left patients with unexpected bills.
A post on X by Lauren Blanchard reported that AI-generated deepfake videos purporting to show American 'Pan-Africanists' expressing support for the junta in Burkina Faso were circulating online. The post suggests the videos were made using AI and may have been intended to create a false impression of American support for the junta. It is unclear who created or distributed the videos, and no specific harm has been reported.
A 73-year-old retired man in Kozhikode, Kerala, was duped into transferring ₹40,000 after receiving WhatsApp messages and a video call that appeared to show a former colleague. Police believe the caller used deepfake technology to mimic the colleague's face and voice. The man realised the fraud when the caller asked for more money and he contacted his actual colleague. Police froze the money in a Maharashtra bank account and said the scam may be part of a wider group operation.
The Lahore safe city project in Punjab, Pakistan, has faced major challenges since 2016, with nearly 1,000 facial recognition cameras out of order. The Punjab Safe Cities Authority (PSCA) blames the contractor for failing to meet contractual clauses. Alternative measures, including reliance on privately-installed cameras, have been taken.
A study by Stanford researchers found that seven popular AI text detectors wrongly flagged over half of essays written by non-native English speakers as AI-generated. The detectors assess text perplexity, and non-native speakers' simpler word choices lead to false positives. The researchers warn that this bias could have serious implications for students and job applicants, potentially leading to discrimination.
Daniel Handford received a scam that used AI to appear more convincing. He tweeted about it to Martin Lewis, who said he had seen similar scams. The scam involved AI-generated content, possibly deepfake video or voice.
An analysis by Bloomberg of over 5,000 images generated by Stability AI's Stable Diffusion found that the text-to-image model amplifies racial and gender stereotypes. The model overrepresented lighter-skinned men in high-paying jobs and darker-skinned people in low-paying jobs, and underrepresented women in positions of power. Stability AI acknowledged the inherent biases in its models and stated it is working on mitigation.
The Dutch Ministry of Foreign Affairs used a secret algorithm called IOB to score short-stay visa applicants based on nationality, gender, and age. Applicants flagged as high risk were automatically moved to an intensive track with higher rejection rates and delays. The ministry's own data protection officer warned of potential ethnic discrimination, but the system continued to be used. The algorithm has profiled millions of applicants since 2015.
Kerala's new AI traffic camera system wrongly recorded a motorcyclist's speed as 1240 km/h and issued a speeding challan. An investigation found the rider had not violated any rules. The system has also been misreading screws on number plates as zeros. Authorities have added human reviewers to control rooms to prevent further errors.
Marvel Studios used AI to generate the opening titles for its Disney+ series Secret Invasion, rather than hiring human artists. The decision was confirmed by executive producer Ali Selim, who said the AI vendor Method Studios created the sequence based on themes and words. The move angered fans and artists, who called it exploitative and morally questionable.
A content creator on Douyin used deepfake technology to pretend to be a Chechen soldier in Ukraine, posting fabricated videos for months. The account gained nearly 400,000 followers and sold items via e-commerce. Users discovered the IP address was in Henan, China, revealing the deception. Douyin suspended the account for spreading misinformation and banned it from making money.
A Cruise autonomous vehicle appeared to block emergency crews responding to a mass shooting in San Francisco's Mission District on 9 June 2023. A police officer in video footage says the robotaxi was blocking emergency medical and fire vehicles. Cruise denied obstructing access and said an employee moved the vehicle within half an hour. San Francisco police said they were investigating.
Each incident is written from the reporting or the account it cites. A report is an allegation, not a finding.
