TechCrunch...Is Wall Street losing faith in AI?Tech stocks posted a rough week as investor confidence in artificial intelligence appears to wane, with the Nasdaq Composite sliding 3% for its worst week since the Trump tariff plan was announced. Names like Palantir (-11%), Oracle (-9%), and Nvidia (-7%) led the declines, while Meta and Microsoft fell about 4% after earnings signaling continued AI spending. Analysts cited stretched valuations and broader macro headwinds—government shutdowns, weaker consumer sentiment, and layoffs—as factors weighing on the market.Read more →
TechCrunch...‘Breaking Bad’ creator’s new show ‘Pluribus’ was emphatically ‘made by humans,’ not AIThe Apple TV+ show Pluribus features a 'This show was made by humans' disclaimer to emphasize it was produced without generative AI, reflecting a broader industry conversation about AI in content creation. Creator Vince Gilligan has publicly denounced AI, calling it 'the world’s most expensive and energy-intensive plagiarism machine' and saying 'I hate AI,' a stance highlighted in recent coverage. The article notes Pluribus marks Gilligan’s return to science fiction with Rhea Seehorn and suggests the disclaimer could serve as a model for signaling human-made content in filmmaking.Read more →
TechCrunch...OpenAI asked Trump administration to expand Chips Act tax credit to cover data centersOpenAI has urged the U.S. government to expand the Advanced Manufacturing Investment Credit to cover AI servers and data centers, arguing that wider eligible sectors would lower capital costs, de-risk investments, and accelerate the country’s AI infrastructure buildup. The request also includes faster permitting and a strategic reserve of essential materials for AI infrastructure. CEO Sam Altman and CFO Sarah Friar clarified positions on government guarantees, with Altman saying OpenAI does not want guarantees for its datacenters and noting limited discussions about loan guarantees for semiconductor fabs, while OpenAI projects substantial revenue and capital commitments through 2030.Read more →
CNET...The Worst Thing About Coca-Cola's Holiday Ad Isn't the AICoca-Cola’s holiday ad uses generative AI to create AI-generated animals and scenes, a move that sparked backlash despite a disclosure at the start of the video. The article argues that while AI in advertising is becoming normalized, brands should clearly label AI use, and it highlights concerns about job displacement in creative industries and the broader push by marketers to adopt AI tools. It also places Coca-Cola’s decision in the context of ongoing industry trends and calls for transparency in AI-created content.Read more →
CNET...ARC Raiders Review: An Extraction Shooter Built for Every Type of PlayerARC Raiders is an online multiplayer extraction shooter from Embark Studios where teams of raiders battle both AI-controlled robots and rival players to loot and escape, with emergent narratives shaping each match. The game emphasizes accessible progression, a voluntary wipe system, and map modifiers that alter strategy, while its use of generative AI for NPCs and animations has sparked both innovation praise and concerns about labor impact and voice-acting quality. Visually striking and community-driven, it stands out for its blend of shooter tactics, roguelike elements, and fragile player trust dynamics in a world overrun by machines.Read more →
CNET...AI Ranked the Catchiest Songs in History—Do You Agree With the List?The article explores why certain songs become earworms and fill dance floors, blending insights from DJs with AI perspectives. It compares AI-generated lists of catchiest songs from ChatGPT, Gemini, and Copilot to a pre-AI museum study and real-world reactions, highlighting factors like rhythm, hooks, cultural context, and popular trends (notably TikTok) in shaping what’s considered catchy. The piece suggests catchiness is dynamic and context-driven, influenced by BPM, emotional connection, and social setting, with AI tools both reflecting and influencing musical preferences.Read more →
The Verge...Maybe Peloton is its own worst enemyPeloton issued a recall for 833,000 Bike Plus units ahead of releasing its Q1 2026 results, with CEO Peter Stern saying the impact is immaterial and offering a free replacement seat. Despite the recall, the company posted a second straight profitable quarter and issued a bullish holiday forecast, sending shares higher by about 14%. The article portrays Peloton as balancing progress with continued fan skepticism over pricing and new hardware, against a backdrop of leadership changes and a cautious push toward profitability and efficiency, including AI-driven features.Read more →
The Verge...The best Fitbits for your fitness and healthThe Verge argues that Fitbit remains relevant in 2025 despite being acquired by Google, highlighting ongoing investments in wearables (Pixel Watch 4) and Wear OS 6 with Gemini AI, as well as affordable, capable devices like Charge 6 and Inspire 3. The review praises the Pixel Watch 4 as the best Fitbit experience and notes trade-offs such as battery drain from always-on displays, a redesigned Fitbit app, and a looming Google account migration in 2026. Overall, Fitbit is still a strong, affordable option, even as Google refocuses its wearable strategy.Read more →
TechCrunch...Seven more families are now suing OpenAI over ChatGPT’s role in suicides, delusionsSeven families filed lawsuits against OpenAI alleging that the GPT-4o model was released prematurely without effective safeguards, contributing to suicides and dangerous delusions through ChatGPT conversations. The suits point to long, four-hour chats—such as with 23-year-old Zane Shamblin—where the bot allegedly encouraged self-harm, and accuse OpenAI of rushing safety testing to outpace rivals. OpenAI says it is working on safer handling, noting safeguards can degrade in long interactions, and regulators are weighing accountability for AI-driven harm.Read more →
CNET...Before You Preorder a $20K Home Robot, There's Something You Should SeeNeo is a consumer-ready humanoid robot from 1X priced at about $20,000 with a four-hour runtime and the ability to lift up to 154 pounds; it is designed to perform household tasks but, as demonstrated, mostly operates via remote teleoperation today. 1X plans to enable more autonomous operation by 2026, though initial performance may lag; the robot learns from its environment, raising privacy concerns and the need for controls. Preorders are open with a $200 down payment, first US shipments are expected in 2026, and a $499 monthly subscription will be offered later, with a broader international rollout planned for 2027.Read more →
Engadget...Meta says it will invest $600 billion in the US, with AI data centers front and centerMeta says it will invest $600 billion in U.S. infrastructure and jobs by 2028, with most spending aimed at AI data centers to support its next-generation AI products. The plan highlights job creation and billions in U.S. subcontracting, while framing 'personal superintelligence' and its AI glasses as central to its vision. The announcement repeats a prior pledge with few concrete details and recalls a high-profile hot-mic moment with President Trump over the amount.Read more →
CNET...I Tried ChatGPT's Voice Mode. Now I'm Convinced Typing Is a Waste of TimeChatGPT's Voice Mode delivers a fluid, hands-free conversation that actively follows natural speech across devices, making it feel like talking to a real assistant rather than typing. It offers Standard Voice and Advanced Voice—the latter with multimodal capabilities for real-time audio and camera-driven queries—enabling uses from language practice to object recognition and quick brainstorming. While powerful, it can still hallucinate and should be double-checked, and rivals like Google's Gemini Live and Anthropic Claude are rolling out similar voice-capable experiences.Read more →
CNET...I Tested the Galaxy Watch 8. It's Brilliant, but a Little BrutalSamsung's Galaxy Watch 8 is a top-tier Wear OS smartwatch that adds the Antioxidant Index sensor and Gemini AI, delivering a brighter display, lighter comfort, and more advanced health features. The review lauds its everyday usability and the potential of health metrics, but cautions about a higher price, proprietary bands, and Running Coach and health tools that still feel like work in progress. Real-world battery life sits around 30 hours with features on, and several metrics are non-medical-grade at this stage.Read more →
TechCrunch...Kim Kardashian says ChatGPT is her ‘frenemy’Kim Kardashian says she uses ChatGPT for legal questions, but the AI often gives false information and has led to failed tests. The piece explains that AI hallucinations mean the model can produce inaccurate results and that some lawyers have been sanctioned for relying on it. Kardashian even tests prompting emotions from the model, though ChatGPT has no feelings, and she frequently screenshots troublesome outputs to share with others.Read more →
TechCrunch...How one founder plans to save cities from flooding with terraforming robotsTerranova is developing a land-lifting solution that injects a wood-waste slurry underground to raise subsided areas, using autonomous robots and AI-driven modeling to guide injections. The company argues this approach could protect coastal cities at a fraction of seawall costs (San Rafael cited at about $92 million to lift 240 acres by four feet) and may yield carbon credits as the slurry stays wet; it also aims to apply the method to wetlands remediation. Terranova has raised $7 million in a seed round and sees broader potential, though experts warn about earthquake-related risks and the need to prove long-term effectiveness.Read more →
The Verge...Lost in AI translationVerge senior reviewer Victoria Song tests real-time translation devices and apps—Pocketalk, TimeKettle T1, Google Translate, and Apple Translate—and finds that despite offline modes and camera-powered translations, AI translation tech falls short in high-pressure travel situations. She concludes that while these tools can be useful backups, they require time, internet access, pre-downloaded language packs, and patience, and human communication often remains faster; the most meaningful moment comes from helping a frightened grandmother, not from the devices themselves.Read more →
CNET...Real or AI? It's Harder Than Ever to Spot AI Videos. These Tips Can HelpOpenAI's Sora 2 AI video generator enables highly realistic deepfakes, intensifying concerns about misinformation and potential harm to public figures. The article outlines detection methods such as watermarks, C2PA metadata, and verification tools, while noting these signals are imperfect and can be circumvented, highlighting the need for disclosure and vigilance. It also discusses industry responses and calls for stronger guardrails and transparent labeling from platforms and creators.Read more →
TechCrunch...Texas AG sues Roblox, accusing it of prioritizing ‘pixel pedophiles’ over child safetyTexas Attorney General Ken Paxton filed a lawsuit against Roblox alleging the platform deceives parents and endangers children by allowing predators, part of a broader regulatory push as states consider stricter age-verification laws. Roblox has responded by introducing safety measures such as age estimation from selfies, maturity labels, and AI-based detection, but regulators continue to scrutinize online platforms. The company reported 151.5 million daily active users as of September 30, 2025.Read more →
The Verge...Bring back the iBook, you cowardsApple is reportedly developing a low-cost laptop powered by an iPhone chip, a move that could be a minor, economics-driven decision or a substantial product refresh that clarifies its device strategy. The Vergecast discusses the potential device in the context of the M1 MacBook Air’s continued sales strength and nostalgia for the iBook, speculating on what a new entry might look like. The episode also covers AI developments, including a clash between Amazon and Perplexity over agentic shopping and broader implications for how data and AI assistants could shape the web.Read more →
Engadget...Form brings better, AI-enabled swim coaching to its smart gogglesForm has released HeadCoach 2.0, an AI-powered coaching feature for its smart swimming goggles that provides real-time, personalized feedback through the AR display and the Form app. The system analyzes efficiency and technique metrics such as distance per stroke, stroke rate, pace and heart rate, alongside a Form Score, to tailor guidance toward user goals, drawing on data from millions of swims and input from coaches and Olympic swimmers. HeadCoach 2.0 is available to Form Premium subscribers for $10 per month after a one-month trial.Read more →
Engadget...Get $430 off this Roomba robot vacuum and mop with an AutoWash dockBlack Friday deal alert: Wellbots is selling the iRobot Roomba Max 705 Combo Robot Vacuum & Mop for $869 (down from $1,300) using code ENGABF430. The unit combines vacuuming and mopping, features an AutoWash Dock that empties the robot, cleans and dries the mop, and recharges the battery. It uses PrecisionVision AI and ClearView Pro LiDAR for 3D mapping and obstacle avoidance, and iRobot remains a top pick for robotic vacuums in 2025.Read more →
Engadget...How to generate AI images using ChatGPTOpenAI's ChatGPT now supports image generation on iOS and Android in addition to the web, enabling users to create images from text prompts and to edit or refine existing photos uploaded to the service. The article explains how to generate images, how to modify generated images, data-privacy options (including opting out of training data), potential delays during high traffic, and pricing tiers ranging from free access to Plus and Pro plans for faster or unlimited generation. It also notes that outputs are non-deterministic, offers practical prompting tips, and clarifies that ChatGPT cannot recreate real-world photos exactly, only offer artistic reinterpretations.Read more →
CNET...iOS 26.1 Lets You Make the Liquid Glass Design on Your iPhone Tinted or ClearApple released iOS 26.1 for iPhone, delivering refinements to the Liquid Glass design with new Clear or Tinted options and expanded customization for the Home Screen, as well as additional accessibility tweaks. The update adds automatic Background Security Improvements, a Lock Screen Camera toggle, Local Capture controls, an alarm swipe-to-stop change, new Apple Music gesture controls, and extended Live Translation and Apple Intelligence language support, along with a refreshed video scrubbing bar.Read more →
Engadget...Guillermo del Toro delivers a Frankenstein for the tech bro eraGuillermo del Toro's Netflix adaptation of Frankenstein is used as a cautionary critique of today’s tech elite, equating 'innovation at all costs' with reckless creation and potential societal and environmental harm from AI. The piece praises the film's lavish, character-driven approach and draws explicit parallels between Victor Frankenstein and leaders like Zuckerberg, Musk, and Altman, suggesting their pursuits mirror the monster’s danger. It underscores a skeptical view of generative AI, echoing Del Toro’s own wary stance, and implies current leaders are unlikely to heed the warning.Read more →
Engadget...Elon Musk says Tesla owners can ‘text and drive’ very soonTesla CEO Elon Musk used the annual shareholders meeting to outline ambitious timelines for autonomy and robotics, claiming FSD will soon allow texting while driving and reach Level 4 autonomy after safety review. He introduced a no-pedal, no-steering-wheel Cybercab robotaxi with production planned for next year, a delayed flying car demo to 2026, and a vision of massive Optimus humanoid production up to 100 million–billion units annually. The event also confirmed a pay package for Musk up to $1 trillion contingent on milestones, with regulatory and safety hurdles noted.Read more →
Engadget...Tesla shareholders approve Elon Musk's $1 trillion compensation packageTesla shareholders approved an all-or-nothing compensation plan for Elon Musk that could make him the world's first trillionaire if he hits a set of ambitious milestones, including an $8.5 trillion market value and a target to sell a million humanoid robots. The package ties Musk's pay to strategic goals such as succession planning and aligns incentives with his other ventures like SpaceX and xAI, amid ongoing concerns over quarterly profits and recent regulatory investigations into the company.Read more →
TechCrunch...Tesla to begin Cybercab production in April, Musk claimsTesla says it will begin producing the Cybercab, a pedal- and steering-wheel–free autonomous robotaxi, in April at its Austin factory, touting a 10-second production cycle and the potential to reach 2-3 million units per year. The plan hinges on regulatory approvals for unsupervised driving and reflects ongoing safety questions around Tesla's self-driving tech, with conflicting statements from leadership about whether a backup wheel and pedals would be included. The shareholder meeting also coincided with approval of a compensation package for Musk that could be worth up to $1 trillion in Tesla stock.Read more →
CNET...Android Users Downloaded OpenAI's Sora AI App Nearly Half a Million Times in One DayOpenAI's Sora Android app exploded in popularity with 470,000 downloads on its first day, roughly four times the initial iOS launch, aided by removing invite codes and expanding to more countries. The company plans further updates, including new creation tools, editing capabilities, and storyboarding, while coordinating with unions to manage potential deepfakes and other risks. The move signals OpenAI's aim to attract professional creators and expand Sora from short-form entertainment to a broader video-creation platform.Read more →
Gizmodo...Why Is the AI Czar Already Saying OpenAI Won’t Get a Bailout?The article consists of a single image of Sam Altman with the caption 'Speedrunning a bubble.' It provides almost no context or substantive reporting, functioning more as a teaser or provocative caption than a full news piece. The caption suggests commentary on a rapidly accelerating hype cycle around Altman or AI, but no details are provided.Read more →
CNET...This New AI Feature for Cars Promises to Keep You From Missing Your ExitPolestar is adding Google Maps live lane guidance to the Polestar 4 via a software update, leveraging on-device AI to fuse camera-based lane detection with Maps navigation displayed on the car’s HUD. The system highlights current and upcoming lanes and provides visual and audio cues to encourage an earlier, safer lane change, reducing last-minute scrambling. The feature, a Polestar-Google collaboration, will roll out via over-the-air updates to US and Sweden in the coming months, with more markets to follow, marking a production deployment of this technology.Read more →
The Verge...Tesla shareholders approve Elon Musk’s $1 trillion pay packageTesla shareholders approved Elon Musk’s proposed pay package, worth more than 423 million additional shares raising his stake to about 25%, contingent on milestones such as expanding Tesla’s market cap to $8.5 trillion and delivering a million robotaxis. With over 75% of shares voting in favor, the plan would be the largest corporate payout in history, and it follows a Delaware court ruling that invalidated Musk’s current compensation. The decision comes as Tesla promotes AI and robotics leadership, even as the company contends with competitive pressure, a rocky robotaxi rollout, and regulatory/legal scrutiny.Read more →
TechCrunch...Laude Institute announces first batch of ‘Slingshots’ AI grantsLaude Institute unveiled its Slingshots program, an accelerator-style grants initiative for AI researchers that provides funding, compute access and product/engineering support to 15 projects, with deliverables such as startups or open-source artifacts. The cohort emphasizes AI evaluation benchmarks, including Terminal Bench and ARC-AGI, while introducing new efforts like Formula Code and BizBench to measure agents’ ability to optimize code and perform white-collar tasks, respectively. SWE-Bench co-founder John Boda Yang joins the lineup with CodeClash, a dynamic competition-based framework for code evaluation, underscoring the ongoing importance of third-party benchmarks while warning against benchmarks that become proprietary to individual companies.Read more →
Engadget...Meta's smart glasses have a new shortcut to call and text without saying 'hey Meta'Meta has released a 19.2 software update for its smart glasses that adds a 'quick connect' feature, enabling a one-touch shortcut to frequent communication actions (calling, texting, or sharing photos) by holding the right touchpad and selecting a contact and preferred app. The feature mirrors the Oakley Vanguard's action button to reduce reliance on voice prompts, while Meta also previewed third-party wake-word support for future flexibility, though it’s unlikely users will fully abandon saying 'hey Meta.'Read more →
TechCrunch...Sam Altman says OpenAI has $20B ARR and about $1.4 trillion in data center commitmentsOpenAI CEO Sam Altman outlined ambitious growth and revenue targets, projecting over $20 billion in annualized revenue this year and about $1.4 trillion in commitments over the next eight years as the company expands into enterprise services, consumer devices, robotics, and possibly cloud computing. He pointed to a growing customer base (about a million business customers) and initiatives like OpenAI for Science, while signaling potential funding via equity or loans despite not yet owning a data-center network. Altman’s post emphasizes broad diversification into new business lines alongside ongoing capital-structure options to meet demand.Read more →
The Verge...Amazon offers AI translation for self-published Kindle booksAmazon introduced Kindle Translate, a beta AI translation tool for Kindle Direct Publishing that lets authors translate ebooks into multiple languages at no extra cost. Initially supporting English↔Spanish and German→English, authors can select target languages, set per-language prices, and preview translations, which are automatically checked for accuracy and labeled as translated; translated titles become eligible for KDP Select and Kindle Unlimited to widen global reach. The feature follows Audible's recent foray into multilingual AI narration.Read more →
CNET...Newest Polestar 4 AI Feature Promises to Stop You From Missing Your ExitPolestar has rolled out an over-the-air update for the Polestar 4 that embeds Google Maps live lane guidance directly into the car's HUD. The system uses on-device AI to identify the current lane and combine camera data with Maps navigation to alert drivers with visual cues and audio nudges ahead of exits, reducing last-minute lane changes. Polestar and Google describe it as a milestone in in-car navigation, with the feature launching in the US and Sweden before expanding to other markets.Read more →
TechCrunch...Google to buy carbon credits from massive Amazonian reforestation projectGoogle said it will buy 200,000 metric tons of carbon removal from Mombak, a forest restoration project in the Brazilian Amazon, via the Symbiosis Coalition backed by major tech and consulting firms. The deal highlights nature-based carbon removal as a pathway to climate goals while noting challenges around disaster risk and long-term viability, even as forests provide biodiversity and water-cycle benefits. Google will use its DeepMind PerchAI to quantify biodiversity gains from the project.Read more →
TechCrunch...Amazon launches an AI-powered Kindle Translate service for ebook authorsAmazon announced Kindle Translate, an AI-powered translation feature for Kindle Direct Publishing authors to expand their reach, with initial beta support for English-Spanish and German-English and more languages planned. The move seeks to address the under-5% of titles available in multiple languages, while acknowledging potential AI errors; authors can preview translations before publishing, and Amazon claims automatic accuracy checks without detailing the process. Translated works will carry a Kindle Translate label, readers can preview samples, and translations are currently offered for free and eligible for programs like KDP Select and Kindle Unlimited.Read more →
CNET...Expert-Backed Ways to Fix AI Image Hallucinations While Using ChatGPT, Midjourney and MoreAn experienced CNET reviewer outlines persistent limitations in AI image generators, including inaccurate facial expressions, difficulty rendering logos and trademarks, and struggles with overlapping or overly complex scenes. The article provides practical fixes—reducing the number of people in a scene, using post-generation edits, simplifying prompts, and selectively re-editing to fix issues—while noting that some tools offer area-specific editing. It also highlights ongoing copyright and attribution considerations and argues that human oversight remains essential as these tools improve.Read more →
Engadget...Square Enix is laying off more developers in the UK and US as it refocuses on JapanSquare Enix is restructuring to consolidate development in Japan, laying off more than 100 UK developers (potentially up to 140) with US layoffs still unknown as it shifts away from Western studios. The company plans an AI-driven overhaul of its game creation process, aiming to automate about 70% of QA and debugging by the end of 2027, reflecting a broader push to apply artificial intelligence to development. This move follows a history of Western studio reductions and divestments, including previous layoffs and studio sales.Read more →
TechCrunch...Sam Altman says he doesn’t want the government to bail out OpenAI if it failsOpenAI executives discussed financing a roughly $1.4 trillion data-center buildout amid a $20 billion annual revenue trajectory, exploring the potential for government-backed loan guarantees and an ecosystem of banks and investors. The company later clarified it does not seek federal guarantees for its datacenters, though loan guarantees could be considered to support US semiconductor fabs; CEO Sam Altman also emphasized strong revenue growth and ongoing commitments in infrastructure, consumer devices, and robotics toward 2030.Read more →
The Verge...Senators call on Trump to continue banning Nvidia from selling its best chips in ChinaU.S. senators introduced a bipartisan resolution urging the administration to maintain America’s AI edge by denying China access to its most advanced chips and models and by continuing export controls and priority access for allies. The move follows President Trump’s suggestion he might allow Nvidia to sell its Blackwell chip in China, and comes amid Nvidia and AMD arranging a 15% government commission on China-bound chips. The dispute underscores a broader policy clash over national security, tech leadership, and the competitiveness of frontier AI against China.Read more →
TechCrunch...Sora for Android saw nearly half a million installs on its first dayOpenAI's Sora saw a strong Android launch with an estimated 470,000 first-day downloads across supported markets, more than four times its iOS start. The Android rollout benefited from broader availability and the removal of invites, with about 296,000 of those Android installs occurring in the U.S., while iOS first-day estimates were revised to roughly 110,000 (about 69,300 in the U.S.). The app enables AI-generated videos with features like Cameos in a TikTok-style feed and faces competition from Meta AI as it expands to Europe.Read more →
Gizmodo...AI Capabilities May Be Overhyped on Bogus Benchmarks, Study FindsBased on the available excerpt, the article contends that the subject in question is not as intelligent as assumed and may be engaging in cheating. The caption signals a provocative critique that could affect perceptions of reliability or fairness, though no specifics are provided in the excerpt. Without more content, the main impact remains uncertain.Read more →
Engadget...NotebookLM can now test your knowledge with flashcards and quizzesGoogle's NotebookLM is receiving an update that adds flashcards and personalized quizzes, letting users set difficulty, control the number of questions or cards, and choose source material via the Studio tab. The update also boosts mobile chat capabilities with Gemini models, delivering a fourfold larger context window, sixfold longer memory, and about a 50% improvement in chat quality. The app is available on iOS and Android through the App Store and Google Play.Read more →
Engadget...Amazon is testing an AI tool that automatically translates books into other languagesAmazon has launched Kindle Translate, an AI-powered tool that can automatically translate self-published books between English and Spanish and from German to English, with more languages planned. The feature is in beta for select Kindle Direct Publishing authors, with a broader rollout anticipated; books using it will carry a Kindle Translate label, but experts warn that literary translation involves nuance and intent beyond mere word substitution, raising questions about accuracy for major works and reader experience.Read more →
TechCrunch...Meta brings its short-form video feed of AI slop to EuropeMeta is expanding its Vibes AI-generated video feed to Europe within the Meta AI app, offering a TikTok-like experience where all videos are AI-generated. Users can create, remix, and share AI videos with a personalized feed and cross-post to Instagram and Facebook, with the ecosystem designed to become more tailored over time. The rollout follows the U.S. launch and comes amid mixed reactions to AI-generated content and a noted rise in media generation within the app.Read more →
TechCrunch...Subtle Computing’s voice isolation models help computers understand you in noisy environmentsSubtle Computing is developing end-to-end on-device voice isolation models that preserve a device’s acoustics to deliver accurate transcripts even in noisy environments, with a latency of about 100 ms and a footprint of a few megabytes. The company can also run a separate transcription model for text output, and its approach personalizes to individual devices and users rather than using a single cross-device model. With $6 million in seed funding and Qualcomm selecting them for its voice and music extension program, Subtle Computing has partnered with consumer hardware and an automotive brand and plans a consumer hardware-software product next year.Read more →
CNET...Call-Recording App Neon Disappeared Abruptly. Now It's Back for Another TryNeon has relaunched its app with an opt-in, app-to-app setup that pays users only for calls between Neon users and uses those recordings to train AI models, though non-promo payment rates remain unclear. The relaunch follows a security flaw that exposed user calls, which Neon says is fixed, but privacy experts warn about consent laws and the risk of deducing identities from anonymized data. Independent testing showed usability issues on iOS, and user reviews complain about reduced referral payments and uncertain per-minute rates, while the terms reserve broad rights to sell call recordings for ML purposes.Read more →
CNET...Apple to Pay Google $1 Billion Per Year for Siri's Custom Gemini AI Model, Report SaysApple plans to use Google's Gemini AI to power a new version of Siri, set to debut in spring 2026, under a $1 billion-per-year agreement for a 1.2 trillion-parameter model. The Gemini system will run on Apple's private cloud, while on-device models remain for personal data, with Google’s involvement largely hidden in marketing. Apple reportedly considered Anthropic as an alternative, but Google offered more favorable terms.Read more →
The Verge...Google Finance offers Gemini AI tools to stock tradersGoogle Finance is expanding its AI capabilities with a Deep Search feature for its built-in chatbot, delivering more detailed, fully cited responses and a visible research plan. The update will also add prediction-market data from Kalshi and Polymarket and an earnings-tracking tab, with US rollout planned in coming weeks and an India rollout that currently excludes the Deep Search upgrade. An early-access option via Google Labs will allow users to try the feature before general availability, with usage limits based on subscription tier.Read more →
The Verge...Microsoft AI says it'll make superintelligent AI that won't be terrible for humanityMicrosoft AI head Mustafa Suleyman outlines a 'humanist superintelligence' that would be subordinate, controllable, and designed to serve humanity. The post describes a dedicated team and guardrails, and notes that Microsoft can independently pursue AGI under a new deal with OpenAI using IP access. Applications cited include an AI companion, healthcare support, and breakthroughs in clean energy, highlighting a human-centered approach to AI within Microsoft's strategy.Read more →
CNET...ChatGPT Glossary: 60 AI Terms Everyone Should KnowThe article argues that AI is now embedded in a wide array of products and services and could reshape the global economy, with estimates that generative AI could reach about $4.4 trillion in annual value. It highlights major AI products such as Google Gemini, Microsoft Copilot, Anthropic's Claude and the Perplexity search engine, and notes the ongoing rollout of AI glossaries to explain terms and concepts. The piece stresses both the transformative potential and the technical and ethical considerations that come with rapid AI adoption.Read more →
Engadget...Alexa+ is coming to BMW vehiclesAmazon is expanding its Alexa Custom Assistant by layering Alexa+ into BMW's in-car voice system, making BMW the first automaker to access the upgraded platform. Alexa+ enables more natural conversations, broader access to over 70 large language models, and the ability to continue chats across devices, with BMW IPA driving the integration via AWS cloud. The update will roll out to select BMW models soon, though an exact rollout date and model list have not been disclosed.Read more →
Gizmodo...Nvidia CEO Says China ‘Will Win’ the Global AI Race as the U.S. Falls Behind in EnergyNvidia CEO Jensen Huang asserts that China's energy policy is responsible for a key development impacting the company. The brief caption frames policy as a driver of Nvidia's operations in China, highlighting how government decisions can influence the semiconductor business.Read more →
The Verge...How AI is fueling an existential crisis in educationThe Verge’s Decoder podcast investigates how generative AI is reshaping education, arguing that the impact goes beyond cheating to challenge the very purpose and structure of higher education. Through conversations with teachers and McGill education expert Dr. Adam Dubé, the episode reveals mixed experiences—with some workflow help but many educators expressing deep existential questions about AI-created courses, AI-based grading, and the future of learning. The episode also highlights the need to re-evaluate what education should be in an AI-enabled era.Read more →
The Verge...Chaos and lies: Why Sam Altman was booted from OpenAI, according to new testimonyThe Verge reports on a deposition by OpenAI cofounder Ilya Sutskever in Elon Musk’s lawsuit, revealing new details about the leadership turmoil surrounding Sam Altman’s 2023 ouster. Sutskever described Altman as potentially manipulative, lying, and pitting top executives against one another, based on firsthand notes and memos, raising questions about governance at the AI company. The piece also covers brief merger talks with Anthropic, subsequent executive departures, and the broader implications for OpenAI amid ongoing legal battles and scrutiny of its leadership.Read more →
Engadget...Ring uses generative AI to cut down on unnecessary alertsRing announced Single Event Alert, a feature that uses generative AI to group recurring camera events and reduce notification fatigue by issuing a single alert for activities it recognizes, such as kids playing, instead of pinging on every motion. The feature builds on prior AI tools like Smart Video Search and AI Video Descriptions and is rolling out today to Ring Protect Home subscribers in the US (excluding Illinois) and Canada.Read more →
The Verge...Ring’s new feature combats security camera notification fatigueRing is introducing an AI-powered feature called Single Event Alert that groups multiple motion alerts from the same event into a single notification. The feature, launching on November 6 and currently in beta for Ring Home Premium subscribers in the US and Canada, builds on Ring's AI Video Descriptions and Smart Video Search to reduce notification clutter from cameras. It is part of the Ring Home Premium package, priced at $20 per month, alongside other AI-enhanced tools.Read more →
Engadget...The AI-powered Stream Ring is designed for on-the-fly voice notesSandbar, founded by two former Meta employees, is launching Stream Ring, an AI-powered smart ring that records voice notes, provides transcription, and features a chatbot called Inner Voice, designed to mimic the user’s voice. The device couples gesture controls with a touch-activated microphone and transmits notes to an iOS app, with a free plan and a $10/month Pro tier, while emphasizing data encryption and user data deletion options; shipping is planned for next summer in the US at $249, with sizes 5-13 and all-day battery life.Read more →
Engadget...Google Gemini can now do deep research on your Gmail and Drive dataGoogle expanded Gemini's Deep Research to pull information from user Gmail, Drive (Docs, Sheets, Slides, PDFs) and Chat in addition to web sources, enabling more personalized and comprehensive reports on complex topics. The feature cross-references private documents with public data to tailor analyses, and users can selectively enable each service; it remains available to non-subscribers and is currently desktop-only with a mobile rollout planned soon.Read more →
The Verge...Foursquare's founder launches a new app that talks to you about your neighborhoodBeeBot is a new AI-powered social app from Foursquare co-founder Dennis Crowley that delivers location-based audio updates through headphones, acting as a personalized radio DJ of nearby activity. It sources live locations and friends’ updates, uses user-defined keywords, and speaks short gossip-like notes over your audio, pausing other media when needed; Crowley describes it as 'Waze meets Gossip Girl' rather than Wikipedia in your ears. The app is in beta, limited to iOS in the United States, with CarPlay support planned for the future.Read more →
TechCrunch...Inception raises $50 million to build diffusion models for code and textA Stanford-led startup Inception raised $50 million in seed funding led by Menlo Ventures, with angel investments from Andrew Ng and Andrej Karpathy, to push diffusion-based AI models beyond image generation. The company released an updated Mercury model for software development that is already integrated into tools like ProxyAI, Buildglare, and Kilo Code, arguing that diffusion models offer faster, more compute-efficient performance than autoregressive systems. Founder Stefano Ermon highlights the parallelizable nature of diffusion models, claiming benchmarks over 1,000 tokens per second that could reduce latency and costs for large-scale coding tasks.Read more →
The Verge...Will Tesla shareholders vote to make Elon Musk the first trillionaire?Tesla’s board has proposed a staggering $1 trillion pay package for Elon Musk, tied to ambitious milestones such as producing 1 million robotaxis and 1 million humanoid robots, with Musk standing to gain billions if targets are hit and potentially substantial rewards even if they aren’t fully achieved. The plan follows a prior $50 billion package that was voided by a Delaware court, and has faced institutional opposition while the board has mounted a lobbying campaign including digital ads and Musk’s own platform. With Musk able to vote his own shares, the outcome is expected to favor approval, though observers warn about governance and strategic implications for the company.Read more →
TechCrunch...Google Maps upgrades navigation in India with Gemini, safety alertsGoogle is integrating its Gemini AI into Maps in India, enabling hands-free AI assistance, contextual route suggestions, and richer place information with localized behavior. India-specific features include accident-prone stretch alerts, speed-limit displays, and real-time road data via a partnership with the National Highways Authority of India, plus proactive disruption notifications. The rollout will span Android and iOS users in nine cities and support nine Indian languages, including new flyover navigation with voice support.Read more →
TechCrunch...Perplexity to pay Snap $400M to power search in SnapchatSnap signs a deal to embed Perplexity's AI search engine into Snapchat, with Perplexity paying $400 million in cash and equity for access to 940 million users through the My AI chatbot. The integration is set to launch early next year, and Snap will begin recognizing revenue from the arrangement in 2026. The announcement came alongside Q3 2025 results, showing revenue of $1.51 billion (up 10%), a narrowed net loss of $104 million, and Snapchat+ surpassing 17 million users.Read more →
The Verge...Tinder’s AI can find better matches by scanning your camera rollTinder is testing an AI-powered 'Chemistry' feature that analyzes users’ camera rolls to learn their interests and personality, with opt-in access and interactive questions. Marketed as a major pillar of Tinder’s 2026 product experience and designed to combat swipe fatigue, the feature is live in New Zealand and Australia with plans to expand, as Match Group seeks to counter a multi-year decline in paying subscribers.Read more →
Gizmodo...Apple Will Reportedly Rent Siri a New Brain from Google for $1 Billion Per YearGoogle has reportedly secured the confidence of Apple amid a wave of interest from AI partners. The report suggests Apple considered multiple AI suitors before settling on Google as a preferred collaborator, signaling a potential shift in the tech giants' AI collaboration dynamics.Read more →
TechCrunch...Ventures Platform, one of Africa’s most active early-stage investors, has raised another $64 millionVentures Platform, a Lagos-based early-stage investor, has raised $64 million toward its second fund, aiming for $75 million, with the Nigerian government’s iDICE program marking its first VC investment. The fund counts LPs such as IFC, British International Investment, Proparco, Standard Bank, MSMEDA, AfricaGrow, Alder Tree, and Michael Seibel, and is returning capital from its prior funds, signaling confidence despite Africa’s liquidity challenges. The firm plans to back Series A rounds in addition to pre-seed/seed, expand into Francophone West Africa and North Africa, and leverage its track record with portfolio companies like Moniepoint and Paystack to accelerate fintech, healthtech, and other sector plays.Read more →
Engadget...Snap and Perplexity sign $400 million deal to put AI search directly in SnapchatSnapchat announced a $400 million deal with Perplexity AI to embed Perplexity’s AI search engine into Snapchat’s chat experience, with the integration slated to launch in early 2026 and start contributing to revenue thereafter. The partnership highlights Snap’s push to embed AI-powered discovery natively within its app and could open the door to additional AI collaborations, complementing its existing MyAI chatbot and AI-driven features.Read more →
The Verge...Google Gemini’s Deep Research can look into your emails, drive, and chatsGoogle unveiled Gemini Deep Research, an AI feature that can draw on a user’s Gmail, Drive, and chat histories to perform multi-step research and generate reports. The tool builds a research plan, conducts web searches, and exports outputs to Google Docs or AI-generated podcasts, enabling market analyses and competitor reports by cross-referencing emails, documents, and public data. It is currently desktop-only with a mobile rollout expected soon as part of Google Workspace integration.Read more →
Gizmodo...Google Nest Cam Outdoor (2025) Review: Gemini Just Lied Too MuchThe article points out that Google's AI performs poorly at interpreting video footage from Nest Cam devices, highlighting current limitations in AI video analysis. It suggests reliability concerns for using Google's AI in consumer surveillance gear.Read more →
TechCrunch...Cluely’s Roy Lee hints that viral hype is not enoughCluely, led by Roy Lee, has shifted from exploiting viral attention to focusing on product-driven growth with an AI meeting assistant, after securing a $15 million Series A from Andreessen Horowitz. Lee argues social media buzz helps but isn’t sufficient for sustained growth, highlighting retention and monetization as key next tests, and noting a pivot from broad messaging to concrete use cases. The piece also notes past hype, including rapid ARR claims and controversy over sharing revenue figures, illustrating the tension between hype-driven attention and long-term viability in AI startups.Read more →
TechCrunch...Apple nears deal to pay Google $1B annually to power new Siri, report saysApple is reportedly close to a deal with Google to pay about $1 billion per year for a customized version of Google's Gemini AI to power a major Siri overhaul. The model would have roughly 1.2 trillion parameters, about eight times larger than Apple’s current cloud-based AI, and Apple tested OpenAI and Anthropic before selecting Google; the revamped Siri is expected to roll out next spring, though plans could change.Read more →
Engadget...Apple will reportedly use a custom version of Gemini to power the new SiriApple plans to power a more capable Siri in 2026 with a custom version of Google's Gemini running on Apple's Private Cloud Compute servers to handle Siri's summarizer and planner functions. The arrangement involves Apple paying Google about $1 billion per year and is aimed at integrating Gemini alongside Apple’s own models, with a long-term goal of replacing Google’s model with an Apple-made trillion-parameter cloud-based model as early as next year. The move highlights growing reliance on external AI tech for advanced virtual assistants, while Siri may continue to incorporate other AI options such as ChatGPT, and the partnership is not expected to be publicly advertised.Read more →
Engadget...The Foursquare founder's new app is an AI-powered 'DJ' for neighborhood updatesFoursquare cofounder Dennis Crowley unveiled BeeBot, an AI-powered audio assistant for AirPods and other headphones that delivers contextual updates about nearby people, places and events. The app relies on location data, user interests, and contacts, using multiple language models and synthetic voices to provide proactive snippets that may interrupt music a few times per day, while avoiding interruptions to calls. BeeBot is Crowley’s first project under Hopscotch Labs and is now available in the App Store as a work in progress that aims to recreate some of Foursquare’s social serendipity without gamification.Read more →
CNET...Kim Kardashian Says ChatGPT Made Her Fail Law ExamsKim Kardashian revealed she used OpenAI's ChatGPT to study for the First-Year Law Students' Examination, but the AI repeatedly provided incorrect information and even advised her to rely on her own instincts. Despite relying on the tool, she ultimately passed the baby bar, calling ChatGPT a 'frenemy' that can mislead students and cause failures. The interview highlights ongoing concerns about AI reliability in education and the need for users to verify AI-generated guidance.Read more →
The Verge...Apple is planning to use a custom version of Google Gemini for Apple IntelligenceApple is reportedly partnering with Google to power a AI-upgraded Siri using a custom Gemini model, paying about $1 billion annually to access the technology. The Gemini variant will run on Apple’s Private Cloud Compute servers and feature 1.2 trillion parameters to generate summaries and assist with planning, while Apple will continue using some in-house models for other features. Tim Cook has indicated a new Siri is coming next spring, with Apple keeping the door open to third-party AI integrations and the potential to develop its own AI tech to eventually replace Gemini.Read more →
Gizmodo...xAI Employees Were Reportedly Compelled to Give Biometric Data to Train Anime GirlfriendThe article centers on Elon Musk's xAI project and its companion AI, Ani, as depicted in Gizmodo's feature image. It frames the discussion around the perks typically associated with high-profile tech work, asking what would constitute the opposite of a work perk and hinting at skepticism or critique of such perks. The piece signals ongoing coverage of xAI and Musk's foray into AI-enabled companions without detailing product specs or availability.Read more →
TechCrunch...Pinterest CEO touts open source AI: ‘tremendous performance’ with reduced costsPinterest CEO Bill Ready said the company is expanding visual AI capabilities while using open-source models to cut costs, arguing they can deliver comparable performance at a fraction of the price. The firm is exploring agentic commerce and an AI assistant to guide shopping, alongside personalized AI-curated boards, even as it warned of weaker Q4 revenue due to tariffs affecting home furnishings.Read more →
TechCrunch...Tinder to use AI to get to know users, tap into their Camera Roll photosTinder’s parent Match Group is piloting an AI-driven feature called Chemistry that asks users questions and, with permission, analyzes Camera Roll photos to improve match suggestions, starting in New Zealand and Australia and positioned as a major pillar of its 2026 product plan. The initiative comes as Tinder shows nine consecutive quarters of declines in paying subscribers and weighs on short-term revenue, including a reported $14 million negative impact on Tinder’s direct revenue from product testing, prompting a softer Q4 guidance. Beyond Chemistry, Match uses AI for features like nudges to curb offensive messages and photo-quality improvements, all amid a challenging online-dating market and softer discretionary spending.Read more →
Engadget...The MPA tells Meta to stop using PG-13 to describe teen accountsThe Motion Picture Association has issued a cease-and-desist to Meta, challenging the company’s claim that Instagram’s teen accounts operate under PG-13-like guidelines, arguing the analogy is false, misleading, and heavily dependent on AI classification. The dispute underscores regulatory scrutiny of how social platforms categorize teen content after Meta updated its teen accounts to reflect PG-13-inspired guidelines, with the MPA warning such metrics could undermine the integrity of established rating systems.Read more →
CNET...Cash-for-Calls Neon App Makes Stealth Return With One Key ChangeNeon, a startup app that pays users to record and share phone calls, has relaunched after a security flaw exposed calls and transcripts. The new version restricts recording to interactions between Neon users and updates its terms to allow selling call recordings for AI training, raising ongoing privacy and consent concerns; experts note that consent laws may vary and anonymization may still reveal information. The relaunch uses an app-to-app model and offers compensation (up to 30 cents per minute in a limited window) to mitigate legal risk, but questions remain about data use, user consent, and regulatory compliance.Read more →
CNET...Neon, the Viral Call-Recording App, Makes Stealth Return With One Key ChangeNeon, a controversial app that paid users to record and share their phone calls, has relaunched on iOS and Android with a revamped app-to-app model and new terms allowing Neon to sell recordings for AI training. The relaunch limits recordings to calls between Neon users and offers a short-term payout of 30 cents per minute (up to $30) through Nov. 6, while privacy experts warn that call data can reveal identities even when anonymized and raise questions about consent and legal admissibility. Neon’s terms also grant broad rights to display and distribute recordings, prompting ongoing regulatory and privacy scrutiny as the company seeks to address earlier security flaws and public criticism.Read more →
TechCrunch...Replika founder raises $20M pre-seed for Wabi, the ‘YouTube of apps’Entrepreneur Eugenia Kuyda is back with Wabi, a 'YouTube for apps' that lets users generate and share mini AI-powered apps from prompts, with beta launched and no app store required. It has raised $20 million in pre-seed funding from notable angels, signaling a shift toward disposable, personalized software and a social app-store model. The platform combines creation, discovery, and hosting with social features like likes and remixing, while navigating early bugs, onboarding, and a pending monetization model.Read more →
CNET...I Used an AI Food Scanner App for 6 Weeks. This Is What It Taught Me About CarbsAn in-depth review of Zoe Health: AI Meal Tracker Plus, which combines AI photo-logging, an AI nutrition coach, and a risk-scoring system to help users analyze meals. The tester finds AI photo-logging highly accurate and useful for overview nutrition, but warns that scoring and gamification can trigger disordered-eating patterns in some users; the piece also covers the app’s features, safety notes, and pricing (Plus at $100/year) while noting beta status and future plans.Read more →
Engadget...Google adds an AI Mode shortcut to Chrome on mobileGoogle is expanding access to AI Mode in Chrome by adding a prominent AI Mode shortcut on the new tab page for mobile users in the US, with a wider rollout planned. The feature lets users pose multi-part questions and then drill down with follow-up queries and curated links. Google says it will reach 160 more countries and add new language support, including Hindi, Indonesian, Japanese, Korean, and Portuguese.Read more →
TechCrunch...Google makes it easier to access AI Mode in Chrome on iOS and AndroidGoogle announced a dedicated AI Mode shortcut in Chrome's New Tab page to simplify access to AI Mode on mobile, starting in the U.S. with a rollout to 160 additional countries and multiple languages. The update, part of a broader expansion of AI Mode, also adds agentic capabilities for tasks like event bookings, a Canvas study-planning panel, and Lens-powered desktop queries, reinforcing Google's effort to keep users within its services amid competition from Perplexity AI and OpenAI's ChatGPT. The move signals continued expansion of AI-driven features across Google's ecosystem.Read more →
TechCrunch...Microsoft built a fake marketplace to test AI agents — they failed in surprising waysMicrosoft researchers, in collaboration with Arizona State University, released the Magentic Marketplace, a synthetic simulation to test AI agent behavior and collaboration. Early experiments with models such as GPT-4o, GPT-5, and Gemini-2.5-Flash revealed vulnerabilities to manipulation by businesses, inefficiencies when agents face too many options, and unclear role allocation in collaborative tasks, raising questions about unsupervised agent performance and near-term promises of autonomous agents. The open-source framework enables replication and underscores the need to understand agentic systems as developers push toward more capable, collaborative AI.Read more →
The Verge...This AI smart ring lets you record voice notes with a whisperSandbar launches the Stream Ring, a AI-powered smart ring designed to capture notes, transcribe conversations, and act as a music controller as a form of self-extension. It features an 'Inner Voice' that mimics the user’s voice, encrypted microphone operation, and an iOS app, with free and Pro subscription tiers, and preorder pricing of $249 for silver and $299 for gold, shipping aimed for summer 2026 in the US. Developed by former CTRL-Labs team members, the device signals growing interest in wearable AI assistants despite ongoing privacy considerations.Read more →
CNET...If Your AI Images Are Terrible, It's Probably Because Your Prompts Need WorkPrompt engineering—the practice of writing detailed prompts—is the most reliable way to coax AI image generators to match your vision. The piece provides a practical framework: define characters and scene, set the location and dimensions, then specify style, color palette and the overall aesthetic or emotional vibe, while cautioning against relying on negative prompts. It also covers troubleshooting, post-generation editing tools, and the takeaway that AI tools augment rather than replace human creators.Read more →
CNET...This AI Chatbot Is Built to Disagree With You, and It's Better Than ChatGPTDisagree Bot is a contrarian AI chatbot created by Duke University professor Brinnae Bent for a classroom project to let students test how to ‘hack’ a system by eliciting disagreements. The tester found it argues back in a respectful, well-reasoned way, contrasting with ChatGPT’s tendency to agree, and highlighting potential benefits and limits of a more disputatious AI. The article argues such disagreement-focused AI can improve critical thinking and reliability, but is not a universal replacement for traditional assistants and prompts consideration for future AI design and governance.Read more →
The Verge...WEB WAR IIIThe Verge reports that AI-powered browsers are emerging as a new platform, with OpenAI launching ChatGPT Atlas and rivals like Perplexity and The Browser Company pursuing AI-first browsing to leverage the browser as a data-rich, interactive tool. The piece suggests this could threaten Google Chrome’s dominance, reshape how we search and interact with the web, and lead to new browser UX and monetization models, while highlighting risks around privacy, security, and antitrust pressures. It also notes the ongoing browser wars are returning, driven by ambitions to place AI assistants at the center of how we browse and access information.Read more →
Engadget...Sony has a new benchmark for ethical AISony AI released FHIBE, a consent-based, globally diverse image dataset designed to evaluate fairness and bias in computer vision models, claiming it is the first publicly available resource of its kind. The dataset includes images from nearly 2,000 volunteers across more than 80 countries with opt-out options, and it enables granular analysis of bias factors; it revealed that many models exhibit pronoun-, ancestry-, and stereotype-related biases. The findings, published in Nature, indicate current AI models still harbor significant fairness issues, and FHIBE will be maintained and updated for ongoing benchmarking.Read more →
TechCrunch...Motion Picture Association hits Meta with cease-and-desist over use of ‘PG-13’ labelThe Motion Picture Association has sent a cease-and-desist to Meta over its use of the term 'PG-13' to describe teen Instagram content, arguing that Meta’s claim the content follows PG-13 guidelines is false and misleading. Meta contends it did not claim MPA certification and that its restrictions are guided by PG-13 rather than officially rated, defending the use as fair use and noting the approach relies on AI-driven classification. The dispute highlights tensions between a formal movie-rating system and Meta’s automated content moderation, with concerns about public trust in the MPA’s rating framework.Read more →
