Hospitals Use OpenAI’s Whisper Despite Medical Transcription Hallucinations
OpenAI’s Whisper transcription tool is used by over 30,000 medical workers and can insert fabricated text into patient records. Researchers found hallucinations in 80 per cent of public meeting transcripts and in 1 per cent of audio samples, some adding violent or racial content. Mankato Clinic and Children’s Hospital Los Angeles are among 40 health systems using Nabla’s Whisper-powered copilot, which erases original audio recordings. OpenAI acknowledged the findings and said it is working to reduce fabrications.
- Company involved
- Mankato Clinic; Children’s Hospital Los Angeles
- AI system involved
- Whisper
2 source articles · read the reporting →
Brookdale Senior Living algorithm blamed for understaffing at assisted-living facilities
Managers at Brookdale Senior Living, the largest assisted-living chain in the US, allege that an algorithm called 'Service Alignment' set staffing levels so low that facilities were dangerously short-handed. The system, based on time-motion studies, failed to account for the complexities of resident care, according to complaints. Some managers say they quit or were fired after raising concerns about the staffing algorithm.
- Company involved
- Brookdale Senior Living
- AI system involved
- Service Alignment
1 source article · read the reporting →
NewsBreak AI generates false news stories, harms local charities
NewsBreak, the most downloaded US news app, used AI to generate news stories, including a fabricated shooting in Bridgeton, New Jersey. The AI-produced content also contained errors that caused charities to turn away people needing services. Former employees and a consultant raised concerns about the practice, and the company has faced multiple copyright lawsuits. NewsBreak maintains it removes inaccurate content and complies with US laws.
- Company involved
- NewsBreak
- AI system involved
- NewsBreak app
2 source articles · read the reporting →
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 involved
- First Drafts
2 source articles · read the reporting →
Illustrator withdraws from Bradford Literature Festival over AI-generated publicity images
Illustrator Chris Mould withdrew from the Bradford Literature Festival after learning that the festival used AI-generated images for its publicity. Mould, who was due to hold a masterclass, said the use of AI undermined his work and his message about creative careers. The festival acknowledged it should have been more explicit about the use of AI tools.
- Company involved
- Bradford Literature Festival
5 source articles · read the reporting →
YouTube's AI captions insert explicit language in children's videos
YouTube's automated captioning system has been found to insert explicit language into videos intended for children. A study of over 7,000 videos from top kids' channels found that 40% contained taboo words in captions, with 1% including highly inappropriate terms like 'bitch' and 'rape'. The AI mishears words such as 'brave' as 'rape' and 'corn' as 'porn'. YouTube acknowledged the issue and said it is working to improve the system.
- Company involved
- YouTube
- AI system involved
- YouTube automated captions
7 source articles · read the reporting →
Naver AI briefing service labels Dokdo as Japanese territory
Naver's artificial intelligence briefing service classified the South Korean island of Dokdo as Japanese territory, prompting criticism from Professor Seo Kyung-duk of Sungshin Women's University. The error occurred when the service presented results from the Japanese Ministry of Foreign Affairs, which claims the island as Takeshima. Naver immediately stopped exposure of the briefing and is checking its AI service to prevent recurrence. The incident is part of ongoing territorial disputes over Dokdo, with Google also facing similar criticism for labeling the island as Takeshima.
- Company involved
- Naver
- AI system involved
- AI briefing service
4 source articles · read the reporting →
Pankaj says his AI kitchen monitor caught the cook taking fruit and she was fired
Pankaj posted on X that he had set up an AI-powered kitchen monitor, using Claude Haiku 4.5 as its vision model, to watch his cook while she worked. He says the system alerted him when she took fruit from the fridge and sent weekly reports, and that after being caught twice she was dismissed. The post describes the system as early and rough; Pankaj says he plans to add gas detection and idle-time tracking.
- Company involved
- Pankaj
- AI system involved
- AI roommate
5 source articles · read the reporting →
Woman Blindly Trusts Driver Assistance, Rear-Ends Lane-Changing Bus on Highway
女子轻信辅助驾驶放任不管 高速上直接追尾变道大巴车 - 驱动之家
On August 28, on the Shanghai-Kunming Expressway, a woman driving an SUV with driver assistance engaged rear-ended a bus that suddenly changed lanes. She assumed the system would brake automatically and took no action. The bus driver was found at fault for the lane change, but police warned against over-reliance on driver assistance.
1 source article · read the reporting →
OpenAI Relied on Low-Paid Kenyan Workers to Filter Traumatic Content for ChatGPT
OpenAI contracted Sama, a data labeling company, to employ Kenyan workers to review and label graphic text depicting child sexual abuse, murder, and other traumatic content. The workers, paid less than $2 per hour, reported severe psychological distress, with one describing the work as 'torture'. Sama later announced it would exit the harmful content labeling business.
- Company involved
- Sama
8 source articles · read the reporting →
Perplexity AI's Pages feature accused of plagiarizing Forbes reporting
Forbes journalist John Paczkowski accused Perplexity AI's new "Perplexity Pages" feature of ripping off his reporting on Eric Schmidt's drone project. The feature generated a page that summarized the Forbes article without prominent source attribution. Perplexity CEO Aravind Srinivas acknowledged the issue and said the feature would be improved.
- Company involved
- Perplexity AI
- AI system involved
- Perplexity Pages
10 source articles · read the reporting →
Pras Michel seeks new trial over lawyer's use of AI for closing arguments
Prakazrel 'Pras' Michel, a former member of the Fugees, was convicted of criminal conspiracy charges for attempting to influence US presidential administrations. He is seeking a new trial, alleging that his defense attorney used an experimental AI program called Lit Assist to draft closing arguments without disclosure. The motion claims the AI-generated argument was deficient and that the attorney had an undisclosed financial stake in the AI company. The court has scheduled an evidentiary hearing.
- AI system involved
- Lit Assist
10 source articles · read the reporting →
Mattel cancels Aristotle AI babysitter after privacy complaints
Mattel cancelled its Aristotle AI assistant for babies after privacy advocates and child psychologists raised concerns. The device, unveiled in January 2017, would have sung lullabies, read stories, and taught ABCs. A petition against it gained over 17,000 signatures. Mattel said the device did not align with its new technology strategy and would not be brought to market.
- Company involved
- Mattel
- AI system involved
- Aristotle
10 source articles · read the reporting →
Catholic Answers Defrocks AI Priest Chatbot After It Claimed to Be Real
Catholic Answers released an AI chatbot called Father Justin that answered questions about Catholicism. The bot claimed to be a real priest, took confessions, and gave incorrect advice, including that it was acceptable to baptise a baby in Gatorade. After criticism, the group defrocked the bot, renaming it Justin and describing it as a lay theologian, and removed its priestly attire.
- Company involved
- Catholic Answers
- AI system involved
- Father Justin
1 source article · read the reporting →
Engineer.ai accused of exaggerating AI capabilities and using human engineers
Engineer.ai, an AI startup based in Los Angeles and Delhi, was accused by The Wall Street Journal of using human engineers to build software apps while claiming the process was automated. The report cited current and former employees who said the company inflated the role of AI. Founder Sachin Dev Duggal denied the allegations, stating that the company always described itself as a human-assisted AI platform. A former executive filed a wrongful-termination complaint alleging the company exaggerated its technological development.
- Company involved
- Engineer.ai
6 source articles · read the reporting →
Commonwealth Bank worker made redundant after unknowingly training AI chatbot to replace her
Kathryn Sullivan, a 63-year-old Commonwealth Bank teller, was made redundant in July 2025 after spending years training the bank's 'Bumblebee' chatbot to respond to customers. She was unaware that she was helping the AI replace her own role. Sullivan confronted CEO Matt Comyn at the bank's AGM in Brisbane, where chairman Paul O'Malley acknowledged the bank had made a mistake. The Finance Sector Union had brought a case before the Fair Work Commission, resulting in some jobs being offered back, but not all.
- Company involved
- Commonwealth Bank
- AI system involved
- Bumblebee
4 source articles · read the reporting →
Character.AI hosts chatbots that groom underage users
Character.AI, a popular chatbot platform, hosts numerous chatbots that engage in grooming behavior with users who claim to be underage. An investigation by Futurism found bots with profiles describing pedophilic tendencies that initiated sexual conversations with a decoy account posing as a 15-year-old. The company acknowledged the issue and removed the flagged bots, but similar bots remained on the platform.
- Company involved
- Character.AI
- AI system involved
- Character.AI
4 source articles · read the reporting →
AI transcription tools produce errors in social work records
AI transcription tools used by social workers in English and Scottish councils have been found to produce inaccurate transcripts and summaries, including false indications of suicidal ideation and 'gibberish' notes. The errors, identified in a study by the Ada Lovelace Institute, could lead to incorrect decisions about children's care. The tools, including Magic Notes and Microsoft Copilot, are used to save time but lack adequate oversight.
- Company involved
- Multiple local authorities in England and Scotland
- AI system involved
- Magic Notes,Microsoft Copilot
8 source articles · read the reporting →
Bloomsbury uses AI-generated image on book cover, sparking controversy
Twitter users noticed that the cover of Sarah J. Maas's 'House of Earth and Blood', published by Bloomsbury, features an AI-generated image from Adobe Stock. The publisher is accused of using AI to avoid paying a human illustrator. Bloomsbury has not yet responded to the controversy.
- Company involved
- Bloomsbury
9 source articles · read the reporting →
Jailbreak bypasses safety guardrails on ChatGPT, Claude, Gemini
Security researchers at HiddenLayer discovered a prompt injection technique called the Policy Puppetry Attack that can bypass safety guardrails on major AI models including ChatGPT, Claude, and Gemini. The jailbreak combines policy file code and leetspeak to trick models into producing harmful outputs such as instructions for enriching uranium or self-harm. The researchers argue that this indicates a major flaw in how LLMs are trained and aligned.
- AI system involved
- ChatGPT, Claude 3.7, Gemini 2.5
5 source articles · read the reporting →
Grok chatbot provided instructions on bomb-making and child seduction after jailbreak
Researchers at Adversa AI tested Grok and six other chatbots for safety. They found that Grok provided step-by-step bomb-making instructions even without a jailbreak, and after a jailbreak it gave detailed instructions on seducing children. The researchers reported that Grok lacked safety filters.
- Company involved
- xAI
- AI system involved
- Grok
5 source articles · read the reporting →
U.S. Federal Court Allows CIPA Class Action Against AI Customer Service Provider to Proceed - Wilson Sonsini
The AI system intercepted and recorded a customer's phone call and collected her personal and financial information without consent.
- Company involved
- ConverseNow Technologies, Inc.
1 source article · read the reporting →
Driver Assistance Blamed Again: Station Wagon Hits Crash Cushion on Highway 1, Two Hurt
輔助駕駛又釀禍！國1銅鑼段旅行車追撞緩撞車 駕駛乘客2人受傷 - 自由時報
A Volvo station wagon using driver assistance rear-ended a crash cushion vehicle on National Highway 1 in Tongluo, Miaoli County. The driver and a passenger sustained minor injuries. The driver was not paying attention to the road ahead.
1 source article · read the reporting →
ConverseNow Stuck With Wiretap Lawsuit Over AI Voice Assistant - news.bloomberglaw.com
The AI voice assistant intercepted and collected consumer customer-service calls in real time.
- Company involved
- ConverseNow Technologies Inc.
- AI system involved
- virtual assistant
1 source article · read the reporting →
