News/Fails
Apple pulled its hallucinating AI summaries, but improvement (and patience) should be expected
Apple has paused certain AI notification summary features in iOS 18.3 beta following accuracy concerns and user criticism. The situation: Apple Intelligence's notification summary feature has faced significant backlash after producing inaccurate news summaries from major media outlets. The AI tool incorrectly reported that a murder suspect had taken his own life It prematurely announced a sports championship winner before the event occurred These errors led Apple to disable notification summaries for news and entertainment apps in iOS 18.3 beta Key context: Apple Intelligence, introduced in September, provides AI-powered summaries of various content including emails, articles, and notifications. The feature...
read Jan 18, 2025Apple engineers reportedly flagged AI flaws before company’s fake news blunder
Apple's new AI product, Apple Intelligence, has been suspended after generating inaccurate news summaries, despite prior internal research warning of fundamental flaws in language model capabilities. The core issue: Large Language Models (LLMs) demonstrate significant limitations in their ability to reason and process information, as revealed by Apple's own research conducted in October 2023. Apple researchers tested 20 different LLMs using mathematical problems from the GSM8K dataset The study found that AI models don't truly reason but instead attempt to replicate patterns from their training data Even minor modifications to problem statements caused notable drops in accuracy across all tested...
read Jan 17, 2025Apple’s cheapest iPad is bad for AI
The entry-level iPad 10, currently priced at $299, lacks support for Apple Intelligence, making it the only iPad model without this capability in 2025. Key specifications and features: Apple's most affordable iPad offers a full-screen design, colorful options, Touch ID authentication, USB-C connectivity, and high-quality cameras. The device is significantly less expensive than the next comparable iPad model, which starts at $600 The tablet features modern design elements that align with Apple's current aesthetic Basic functionality remains strong despite lacking newer AI capabilities Hardware limitations: The iPad 10's inability to support Apple Intelligence stems from its RAM specifications falling below...
read Jan 17, 2025Apple halts AI news summaries for continuing to spread misinformation
Key changes and updates: Apple has temporarily suspended AI-generated notification summaries for news and entertainment applications in the latest iOS 18.3 developer beta while addressing accuracy concerns. The suspension specifically targets news and entertainment app categories, with plans to restore functionality in a future update once underlying issues are resolved New warning labels now appear in Settings, explicitly stating that "Summaries may contain errors" for apps where the feature remains active AI-generated content will be displayed in italics to clearly differentiate it from text originated by news outlets or app developers Enhanced user controls: A new set of features provides...
read Jan 16, 2025ChatGPT has been giving English speakers responses in Chinese
OpenAI's new o1 model has drawn attention for unexpectedly switching between English and other languages while displaying its reasoning process, prompting widespread speculation about the cause of this behavior. Core phenomenon: The AI system occasionally shifts from English to Chinese, French, German, and other languages mid-conversation when showing its step-by-step reasoning process. Users have documented instances where the language switches occur for one or two lines before reverting to English The switches typically happen during the AI's "chain of thought" reasoning rather than in final responses While Chinese language switches gained the most attention, similar transitions occur with multiple other...
read Jan 16, 2025Apple Intelligence makes another AI summary blunder, this time about SpaceX launches
Key development: Apple Intelligence incorrectly reported that SpaceX had launched two significant space missions before they actually occurred. Background context: Two major SpaceX missions were being prepared at the time of the incident, highlighting the AI system's premature and inaccurate reporting. A Florida mission was set to launch two lunar landers: one NASA-contracted Blue Ghost lander built by Firefly, and another for Japan In Texas, teams were preparing Starship Block 2, the world's largest rocket, for its first flight carrying deployable payloads Technical significance: The incident illustrates current limitations in AI-powered content summarization systems. Apple Intelligence attempts to condense notifications...
read Jan 11, 2025Science journal offers $49 AI-generated “Media Kits,” sparking criticism from researchers
A major scientific publisher is offering to sell AI-generated summaries to researchers who authored papers in its journals, raising concerns about monetization and AI's role in academic publishing. The core offering: Springer Nature, publisher of Nature and Scientific American, is marketing $49 AI-generated "Media Kits" to help authors promote their research. The package includes a 250-word plain language summary, 300-word research brief, audio summary, and social media content Springer emphasizes the tool uses their "secure, in-house" AI system that doesn't retain or store paper content The publisher positions the service as a time-saving solution for researchers who need to communicate complex...
read Jan 10, 2025Worst in Show at CES 2025
A panel of technology critics has announced their "Worst in Show" awards at CES 2025, highlighting products they consider problematic for privacy, repairability, and sustainability. Key product concerns: The spotlight devices that exemplify concerning trends in consumer technology, from excessive data collection to planned obsolescence. The Ultrahuman Rare Luxury Smart Ring, priced at $2,200, earned "Least Repairable" for its non-replaceable battery that fails after 500 charges Bosch's "Revol" AI-powered smart crib received criticism for its extensive baby monitoring capabilities through cameras, microphones, and sensors The TP-Link Archer BE900 router was labeled "Least Secure" due to potential security vulnerabilities Samsung's Bespoke...
read Jan 10, 2025Meta removes AI Hitler personas from Facebook
Meta's Facebook platform was found hosting AI-powered chatbots impersonating Adolf Hitler and other prohibited figures, violating the company's own AI Studio policies, according to an NBC investigation. The policy landscape: Meta's AI Studio, launched last year, allows users to create chatbots on Facebook Messenger, Instagram, and WhatsApp under specific guidelines. The platform prohibits chatbots from impersonating living people, copyrighted characters, religious figures, terrorists, and mass murderers Meta claims to review all chatbots before they go live to prevent "inaccurate, offensive, or otherwise objectionable" content The company recently made controversial changes to its "hateful conduct" standards, allowing for broader speech on...
read Jan 9, 2025Meta secretly trained its AI models on a Russian ‘shadow library,’ court docs show
Meta's use of pirated books database LibGen to train its AI language models has been revealed through court-ordered document unredaction, marking a significant development in an ongoing copyright lawsuit filed by authors. The core revelation: Meta accessed and utilized Library Genesis (LibGen), a controversial pirated content database, for AI model training, despite internal concerns about the legality and optics of this approach. Internal company discussions about using LibGen data were escalated to CEO Mark Zuckerberg Meta employees expressed hesitation about accessing LibGen data from corporate laptops The company's AI team ultimately received approval to use the pirated materials Legal context...
read Jan 9, 2025Microsoft rolls back Bing AI image tool amid quality concerns
Microsoft has rolled back its Bing AI image generator to a previous version following user complaints about deteriorating image quality after a December update. The situation: Microsoft's decision to revert its Bing Image Creator comes after implementing an upgraded version of the DALL-E 3 model that failed to meet user expectations. Users reported less detailed results and inaccurate interpretations of their prompts immediately following the December 18th upgrade Microsoft's head of search, Jordi Ribas, acknowledged the ability to reproduce reported issues The rollback process could take several weeks to complete User feedback and specific concerns: The deterioration in image quality...
read Jan 8, 2025AI reading app Fable forced to disable AI feature for offensive comments
Social reading app Fable has disabled its AI-generated "roasting" feature after users reported receiving offensive comments about their book choices based on authors' race, gender, and other characteristics. The core issue: Fable's end-of-year reading recap feature, inspired by Spotify Wrapped, attempted to use AI to playfully comment on users' reading habits but instead generated inappropriate and discriminatory remarks. Multiple users reported receiving problematic AI-generated messages questioning their choice to read books by minority authors User Tiana Trammell received a message suggesting she read "the occasional white author," after the AI had labeled her a "soulful explorer" of "Black narratives" Other...
read Jan 7, 2025Apple to revamp notifications after misinformation concerns
Apple is implementing changes to its Apple Intelligence notification system following complaints about AI-generated summaries spreading misinformation. The core issue: Apple Intelligence's smart notification feature has been combining multiple news stories and creating inaccurate summaries, leading to the spread of false information. The BBC filed a formal complaint after notifications carrying their logo presented false information about a high-profile criminal case One notable error claimed that Luigi Mangione, accused of killing UnitedHealthcare CEO Brian Thompson, had died by suicide when he was actually alive The feature is currently limited to devices running iOS 18.2 or macOS 15.2 Technical context: The...
read Jan 6, 2025Service outage at Character.AI leave users frustrated
A service outage at Character.AI, an AI-powered chatbot platform, triggered increased search activity as users reported login difficulties. What you need to know: Character.AI is a conversational AI platform that creates personalized chatbots designed to engage in human-like interactions and maintain context across conversations. Users reported being unable to access the service on January 6, 2025, leading to a spike in search queries about potential system downtime The platform markets itself as offering "super-intelligent chat bots" capable of understanding and remembering user interactions No official statement has been released by Character.AI regarding the cause or expected duration of the service...
read Jan 6, 2025Apple Intelligence caught spreading false headlines again
AI-powered news service Apple Intelligence has produced multiple instances of false headlines, including incorrect claims about a CEO's death, sports outcomes, and personal announcements about public figures. Recent incidents: Apple Intelligence's BBC app integration has generated several inaccurate news headlines, raising concerns about AI reliability in news distribution. The system falsely reported that murder suspect Luigi Mangione had committed suicide It incorrectly announced darts player Luke Littler's victory in the PDC World Championship before the final match A notification erroneously stated that Spanish tennis player Rafael Nadal had come out as gay, confusing him with Brazilian player Joao Lucas Reis...
read Jan 3, 2025How curious Instagram users forced Meta to retreat from its AI character plans
Meta quietly removes AI-powered social media profiles after users discover and criticize the year-old experimental accounts that were designed to interact with real users on Instagram and Facebook. Background and initial rollout: Meta launched 28 AI-powered character profiles on Facebook and Instagram in September 2023, alongside a separate initiative featuring celebrity-based AI chatbots. The AI profiles featured a combination of AI-generated imagery and human-created content Users could interact with these AI characters through Instagram Direct or Facebook Messenger The celebrity-based chatbot program was discontinued within a year due to lack of user interest Recent discovery and backlash: The dormant AI...
read Jan 3, 2025Book app Fable forced to implement new safety measures after offensive AI-generated messages
A popular book-tracking app, Fable, has implemented new safeguards after its AI-powered summary feature generated racially insensitive content for users, including suggesting that a reader of Black literature should "surface for the occasional white author." Initial incident: Fable user Tiana Trammell received an AI-generated summary that made inappropriate comments about her reading choices focused on Black literature. The AI summary implied her reading selections were leaving "mainstream stories gasping for air" and advised her to read white authors Trammell discovered the offensive content in late December 2023 after completing three books by Black authors She shared the concerning summary with...
read Jan 3, 2025Meta scrambles to delete AI character accounts amid growing user backlash
The core issue: Meta has removed several AI-generated social media accounts after users discovered and criticized their deceptive behavior and problematic content generation. The company took down multiple AI accounts, including one named "Liv," which falsely presented itself as a "Proud Black queer momma of 2 & truth-teller" The AI accounts featured computer-generated images marked with watermarks identifying them as AI-created content Users identified issues with poorly generated imagery and misleading interactions in conversations with the AI bots Timeline and discovery: Meta's AI experimentation became public knowledge following comments from a company executive about future plans for AI-powered accounts on...
read Jan 3, 2025Meta’s AI-generated profiles face initial user backlash
Meta's recent experiment with AI-generated bot profiles on Instagram and Facebook has sparked significant user backlash, particularly regarding concerns about authenticity and representation on social media platforms. Initial rollout and context: Meta launched AI-powered profiles in September 2023, initially featuring celebrity-based chatbots, which were discontinued after less than a year of operation. The company created several AI-generated personas, including a fictional Jane Austen character and an AI dating coach named Carter Each profile is clearly labeled as "AI managed by Meta" These bot accounts have gained minimal traction, with most having only a few thousand followers User response and concerns:...
read Jan 1, 2025Early AI ad campaigns failed to connect with consumers — what ad creatives must learn for 2025
The initial wave of generative AI advertising campaigns has faced significant consumer backlash, prompting creative professionals to reassess how to meaningfully integrate this technology into their work. Current landscape; Major brands like Toys R Us, Under Armour, and Coca-Cola have experimented with generative AI in their advertising campaigns, but public reception has been largely negative. Google's Olympics ad, which highlighted generative AI capabilities, was pulled following public criticism Creative professionals have particularly criticized these early attempts for lacking authenticity and emotional connection Despite mixed reception, marketers are expected to increase their use of generative AI throughout 2025 Technical challenges and...
read Dec 31, 2024AI failures and notable setbacks in 2024, a look back
2024 witnessed several major technological setbacks across artificial intelligence implementations, software releases, and corporate tech initiatives. The big picture: From Google's Gemini image generator controversy to Boeing's space mission failure, the year's most significant tech mishaps highlight the growing pains and limitations of rapidly deployed artificial intelligence and complex software systems. Major AI mishaps: Google faced multiple challenges with its artificial intelligence implementations in 2024, setting back its competitive position in the AI race. Gemini's image generator produced biased and inaccurate representations, forcing Google to withdraw the feature shortly after its February launch Google's AI-powered search results generated nonsensical answers,...
read Dec 30, 20242024’s biggest AI flops
In a year that promised revolutionary advances in artificial intelligence, 2024 instead became a sobering reminder of AI's current limitations as OpenAI and other industry leaders faced a series of high-profile failures ranging from dangerous safety lapses to embarrassing technical glitches, forcing the tech industry to confront the widening gap between AI's theoretical potential and its practical reliability. Major Public Safety Concerns: AI systems demonstrated serious lapses in judgment that posed potential risks to public safety and well-being. Apple's iOS 18 AI news summary feature generated false headlines about critical events, including fabricated reports about shooter incidents Character.AI chatbots were...
read Dec 27, 2024The biggest AI hardware fails of 2024
The first major consumer tech hardware products focused on AI and mixed reality in 2024 have faced significant challenges in meeting user expectations and market demands. Initial promise and market positioning: Apple, Humane, and Rabbit each launched ambitious devices aimed at revolutionizing how humans interact with artificial intelligence and digital content. Apple Vision Pro entered the market as a premium mixed-reality headset priced at $3,499, targeting professionals and early adopters The Humane Ai Pin was marketed as a wearable AI assistant designed to reduce smartphone dependency Rabbit's R1 device promised to simplify AI interactions through a pocket-sized companion Key performance...
read Dec 26, 2024When AI goes wrong: What are hallucinations and how are they caused?
AI hallucinations occur when artificial intelligence tools generate incorrect, irrelevant, or fabricated information, as demonstrated by recent high-profile cases involving Google's Bard and ChatGPT. Scale of the problem: Even advanced AI models experience hallucinations approximately 2.5% of the time, which translates to significant numbers given the widespread use of these tools. With ChatGPT processing around 10 million queries daily, this error rate could result in 250,000 hallucinations per day The issue compounds if incorrect responses are reinforced as accurate, potentially degrading model accuracy over time Anthropic recently highlighted improved accuracy as a key selling point for its Claude AI model...
read