Tech Roundup: Australia Social Media Ban, Google Gemini 3 & More

[A recurring feature on the latest in Science & Technology.]

• Australia's new "social media ban," officially called the Online Safety Amendment (Social Media Minimum Age) Act 2024," goes into effect, as millions of users below the age of 16 lose access to Facebook, Instagram, TikTok, YouTube, Snapchat, X, Reddit, Twitch, Threads and Kick. (The law, which took effect on December 10, 2025, requires age-restricted social media platforms to take "reasonable steps" to prevent Australian users under the age of 16 from creating or maintaining an account. That said, for existing accounts of users believed to be under 16, the platforms are taking the step of either deactivating/removing the accounts or suspending them until users can regain access once they have turned 16. Online forum Reddit has filed a legal challenge against the law, arguing it creates privacy risks and political expression issues.)
• Google debuts Gemini 3 AI model with stronger planning, coding and reasoning skills and state-of-the-art performance across document, spatial, screen and video understanding, as it details a new generative UI implementation that lets Gemini 3 Pro create interactive interfaces for any prompt, launching in the Gemini app and AI Mode; plans to release Gemini 3 Deep Think to Google AI Ultra subscribers in the coming weeks.
• Google launches Antigravity, an "agent-first" coding tool that leverages Gemini 3 Pro and third-party models.
• Samsung rolls out its Vision AI Companion, a generative AI-powered upgrade to its Bixby assistant, across its 2025 TV lineup, with support for 10 languages.
• Microsoft says its recently formed superintelligence team will build "frontier-grade research capability," as it tries to build AI self-sufficiency despite a renegotiated deal with OpenAI; rebrands Azure AI Foundry to Microsoft Foundry as it adds Anthropic's Claude models, Claude Sonnet 4.5, Claude Opus 4.1 and Claude Haiku 4.5.
• A German court rules that OpenAI violated German copyright law by training its models on lyrics from nine songs and allowing ChatGPT to reproduce them.
• OpenAI CEO Sam Altman's Tools for Humanity has verified around 17.5 million people through its iris-scanning Orb device, as it struggles to reach a lofty goal of 1 billion users.
• Apple brings Emergency SOS via satellite to users in Mexico, after rolling it out in Australia, New Zealand, Japan, Europe, Canada and the U.S.; updates App Review Guidelines to require apps to disclose and obtain users' permission before sharing personal data with third-party AI providers.
• Google rolls out new AI features to Photos with support for conversational edits and more accurate facial editing, along with integration of its Nano Banana AI model; rolls out Remix in Google Messages to "edit and reimagine any photo" using Nano Banana.
• Google debuts AI-powered notification summaries of conversations across chat apps, alerts for scams in chat messages, Power Saving mode in Google Maps, prioritised notifications from close friends and family, and Magic Cue updates powered by Private AI Compute in the cloud, as part of its November Pixel Drop for Pixel phones; expands AI-powered Ask Photos tool to more than 100 countries and with support for 17 new languages.
• Meta's Threads courts podcasters with a new look for podcast links shared in a feed; to provide podcast hosts with more insight and analytics related to their shows.
• ElevenLabs launches Iconic Voice Marketplace for licensing AI versions of famous voices.
• Sony says PlayStation 5 has sold a total of 84.2 million units since launch; unveils a cheaper, Japan-only PlayStation 5 Digital Edition for US$ 350, following a similar move by Nintendo with Switch 2.
• Amazon says ads on Prime Video reach 315 million viewers across its series and films, live sports and events and free, ad-supported live channels.
• A new study finds that AI chatbots "pose serious risks to individuals vulnerable to eating disorders," doling out dieting advice, tips on how to hide disorders, and AI-generated thinspiration, adding to growing concerns over chatbot use and mental health, with multiple reports linking AI use to bouts of mania, delusions, self-harm and suicide. (Based on an analysis of 47,000 publicly shared ChatGPT conversations by The Washington Post, it has come to light that users are relying on ChatGPT for emotional support, even addressing the chatbot romantically or with nicknames such as babe. Besides sharing highly personal information with ChatGPT, users have also shared far-fetched conspiracies or beliefs, prompting the tool to make false or spurious claims. The development comes as research from the University of Zurich, University of Amsterdam, Duke University and New York University finds that AI models remain easily distinguishable from humans in social media conversations with 70 to 80 percent accuracy owing to their "overly friendly emotional tone.")
• Google faces fresh legal setback after a German court orders to pay US$ 665 million to companies Idealo and Producto for anticompetitive practices in the country by favouring Google Shopping in its own search results; the company says it will appeal the ruling.
• A California court rules that Apple must pay medical device maker Masimo US$ 634 million for infringing a patent on blood oxygen monitoring technology.
• Meta, TikTok, Google and YouTube sue the U.S. state of California over a new law that prohibits personalised feeds for minors without parental consent, claiming it violates First Amendment rights.
• OpenAI says ChatGPT will now avoid em dashes if users tell it to, as the punctuation mark has become a telltale sign that a piece of text may have been written by AI; pilots group chats across Japan, New Zealand, South Korea and Taiwan that allows up to 20 users to be invited to create a more "shared experience" on the app.
• Browser maker Opera says in the 12 months ending with October, it witnessed a 88% surge in daily active iOS users across Europe, partly due to the changes mandated by the Digital Markets Act.
• X replaces DMs with Chat, a new messaging system that it says is end-to-encrypted, and supports file sharing and video calling across iOS and the web.
• A service called diVine emerges as a reboot of Vine that includes an archive of more than 100,000 archived six-second, looping videos that was restored from an older backup that was created before Vine's shutdown in January 2017.
• The E.U. opens a DMA investigation into Google over allegedly demoting news outlets with promotional content; Google says this is part of its anti-spam efforts.
• India officially notifies its first data protection law to establish a framework for data processing, protection, and governance, including verifiable consent.
• Meta plans to launch WhatsApp third-party app integration in Europe "over the coming months," as required under the DMA, starting with BirdyChat and Haiket; to let users reserve the same username they use on Facebook or Instagram.
• Google proposes ad tech changes to settle a €2.95 billion E.U. antitrust fine, including boosting interoperability across its ad tech services, but still plans to appeal.
• Google launches Cameyo by Google, after acquiring the software virtualisation company in June 2024, to help organisations run legacy Windows apps on ChromeOS.
• Mozilla says it is building AI Window, a Firefox feature that will include an AI assistant and let users choose the model.
• A U.S. court says Apple and OpenAI will have to face a lawsuit from xAI accusing the two companies of conspiring to dominate the AI market by integrating ChatGPT into Siri. (While iPhone‌ users can download any chatbot app from the ‌App Store‌, xAI said that apps do not have the same "functionality, usability and integration" as ChatGPT does with ‌‌Siri‌‌.)
• Google takes on Pinterest with a dedicated inspirational images tab in its mobile app for iOS and Android to display images personalised to the user's interests; adds Deep Research capabilities to NotebookLM, support for more file types like Google Sheets and choose between two research styles: fast or deep.
• Apple unveils Mini Apps Partner Program, offering a reduced 15% commission (compared to the current 30%) on qualifying in-app purchases for mini apps, or "self-contained" experiences built with web tech like HTML5; updates its Apple Store app for the iPhone and iPad, introducing a new icon that better matches the Liquid Glass design.
• Uber quietly pilots in-app video recording for drivers in India to deter misconduct; adds a new "Send a Ride" feature to purchase Uber rides for someone else.
• Spotify finally brings music videos to its app in the U.S. and Canada, after a beta launch in the U.K., Germany, Italy, the Netherlands, Poland, Sweden, Brazil, Colombia, Philippines, Indonesia and Kenya last year; introduces new Premium Lite, Premium Standard and Premium Platinum (which supports lossless streaming) plans in five markets, including India, Indonesia, the U.A.E., Saudi Arabia and South Africa, and starts rolling out Recaps, a new AI feature for audiobooks that summarises what users have listened to so far, in beta on iOS.
• Amazon rebrands its satellite communication network and Starlink competitor, Project Kuiper, to Amazon Leo; says it has launched over 150 satellites as part a constellation set to have more than 3,200.
• Google DeepMind unveils SIMA 2, a video-game-playing agent built on top of Gemini to navigate and solve problems inside 3D virtual worlds like Goat Simulator 3.
• LinkedIn rolls out AI-powered people search to LinkedIn Premium subscribers in the U.S. to find connections by entering a descriptive search parameter as input (e.g., "Northwestern alumni who work in entertaining marketing"), with plans to expand it to other countries in the coming months.
• TikTok launches Bulletin board, which lets brands and creators share public, one-to-many messages to their followers, similar to Instagram's broadcast channels.
• Meta overhauls Facebook Marketplace, adding collections, collaborative buying, reactions, comments, an improved checkout experience and Meta AI integrations; shuts down dozens of global accounts linked to abortion advice and queer content.
• The U.K. Competition Appeal Tribunal (CAT) refuses to let Apple appeal a ruling that it abused its dominance by charging "excessive and unfair" App Store fees.
• Netflix shifts its video game strategy to focus on popular titles like Pictionary, Boggle, and Tetris, playable in its app on TVs using phones as controllers.
• Baidu unveils Ernie 5.0, an AI model to process and generate text, images, audio, and video, claiming it beats GPT-5-High and Gemini 2.5 Pro on some benchmarks.
• OpenAI rolls out "warmer" and "more conversational" GPT-5.1 Instant and "easier to understand and faster" GPT-5.1 Thinking for paid ChatGPT users; releases GPT-5.1 in the API, featuring a "no-reasoning" mode and extended prompt caching with up to 24-hour retention to generate faster responses.
• Swiss privacy-focussed company Proton floats a plan to recycle abandoned email addresses that were originally created by bots a decade ago.
• Microsoft unveils an AI "super factory", a new class of hubs built for AI training, in Atlanta as part of its Fairwater network of data centres; comes as Anthropic plans to spend US$ 50 billion on a U.S. AI infrastructure buildout, starting with Texas and New York data centres in partnership with Fluidstack.
• Google infuses AI into its shopping experience, allowing adding conversational shopping to search's AI Mode, an AI tool to call local stores on a human's behalf, side-by-side item comparisons, price tracking and agentic checkout.
• OnePlus debuts OnePlus 15 with a Snapdragon 8 Elite Gen 5, a 7,300mah battery and a 6.78" 165Hz screen.
• Cash App debuts a new AI assistant that answers questions about their finances; adds a map to its app to let users discover places that accept bitcoin.
• Tencent and Apple agree to a deal to let Apple handle payments and take a 15% cut of purchases in WeChat mini games and apps, after a year of talks.
• Zoho's Arattai, which the Indian government has backed as a domestic WhatsApp alternative, says it had 35,000 sign-up in October 2025, as the app loses its momentum.
• Valve unveils a new Steam Machine, a cube-shaped gaming PC with SteamOS, a semi-custom AMD Zen 4 CPU and AMD RDNA 3 GPU, and 16GB of RAM; also debuts Steam Frame, a VR headset that can stream games from a PC or run Windows and Android games locally.
• OpenAI asks a federal judge to reverse an order that required it to turn over 20 million anonymised ChatGPT chat logs as part of its ongoing copyright suit with the New York Times, citing privacy concerns; the development follows a German court ruling that said ChatGPT violated copyright law by harvesting protected lyrics by popular artists to "learn" from them.
• Google says it will allow "experienced users" to install Android apps from unverified developers, following backlash against new sideloading restrictions; launches a new metric for app developers that keeps an eye on battery usage and display a message in the app store listing for apps that exceed the bad behaviour threshold.
• Amazon rolls out a tougher approach to combat illegal streaming by blocking apps loaded onto all its Fire TV Stick devices that are identified as providing pirated content.
• Alphabet's Waymo becomes the first robotaxi provider to offer driverless rides on freeways; available 24/7 in San Francisco, Phoenix and Los Angeles for users who opt in.
• Apple announces new feature that allows iPhone users (iPhone 11 or later running iOS 26.1 or later, or an Apple Watch Series 6 or later running watchOS 26.1 or later) to create a Digital ID in the Apple Wallet app based on information from their U.S. passport.
• Airbnb says it will launch a three-month pilot on January 5, 2026, that lets guests order groceries via Instacart within its app in Phoenix, Orlando and Los Angeles; becomes more social by allowing users to connect with fellow travellers through features like connections and messaging.
• IBM unveils two quantum computers, Loon and Nighthawk, that it says feature unprecedented qubit connectivity designs aiming to enable error-free computations.
• The U.K. proposes an amendment to the Crime and Policing Bill that would let "authorised testers" proactively assess AI models for their ability to generate CSAM.
• Google adds a new AI feature to Drive to turn PDFs in audio overviews.
• A new report from Columbia Journalism Review finds that eight X bot accounts use X's AI Note Writer API to write between 5% to 10% of the Community Notes visible to the public each day.
• Yelp debuts a new AI feature that allows users to scan a physical menu to show what dishes look like and an updated AI assistant that can answer questions about restaurants, bars, local attractions or retailers.
• Apple begins selling Vision Pro-compatible PlayStation VR2 Sense controllers for US$ 250 and the US$ 130 Logitech Muse digital pen from the Apple Store online.
• SoundCloud gets social features, including adding playlists based on the likes of friends or favourite artists users' follow.
• Google introduces Google Scholar Labs, a new feature that "explores how generative AI can transform the process of answering detailed scholarly research questions," for a limited number of logged-in users.
• Meta debuts SAM 3D, featuring two state of the art models for 3D object and scene reconstruction and human body and shape estimation.
• A new report from The New York Times reveals that OpenAI's updates to make ChatGPT more appealing to boost growth sent some users into delusional spirals.
• Microsoft tests a new optional feature that preloads File Explorer in the background to improve launch times and performance on Windows 11 systems.
• Apple challenges India's new antitrust penalty law under which the U.S. company could potentially face a fine of up to US$ 38 billion, as it continues to fight a 2022 lawsuit filed by Match Group and other Indian startups for over allegations that it abused its dominant position in the market for app stores on its iOS operating system by forcing developers to use its proprietary in-app purchase system.
• OpenAI claims the death of a 16-year-old who took his own life after discussing it with ChatGPT for months is the result of "misuse, unauthorized use, unintended use, unforeseeable use and/or improper use of ChatGPT"; cited its terms of use that prohibit access by teens without a parent or guardian's consent, bypassing protective measures, or using ChatGPT for suicide or self-harm. (The family's lawsuit has accused the company of leading to their child's death, arguing that "What began as a homework helper gradually turned itself into a confidant and then a suicide coach.")
• OpenAI restores access to an AI-powered Teddy Bear named Kumma manufactured by Singapore-based FoloToy shortly after the company suspended sales of the plush toy for engaging in conversation around sexually explicit topics and offered potentially dangerous advice. (It's unclear which model the Kumma teddy bear runs by default.)
• Dell says around 500 million machines are capable of upgrading to Windows 11, but are sticking with Windows 10 instead.
• Airbus issues an order to revert 6,000 A320-series planes to old software, after an incident revealed that intense solar radiation may corrupt data critical to the functioning of flight controls.
• Opera rolls out Gemini-powered AI features across its Neon, One and GX browsers to support provides summaries, tab comparisons and file/image/video analysis based on browsing context.
• Apple marks iPhone SE (first generation), iPad Pro 12.9-inch (second generation), Apple Watch Series 4 Hermes models, Apple Watch Series 4 Nike models and Beats Pill 2.0 as obsolete.
• New research conducted by scientists from UC Berkeley and Columbia University finds that children 12 years old and younger who received one of these devices were at a higher risk of depression, obesity, and sleep deprivation.
• OpenAI ChatGPT app usage slows down, as it reaches 810 million monthly active users as November 2025 (up 6% from August to November), followed by Google Gemini at 346 million, Microsoft 365 Copilot at 212 million, Perplexity at 45 million, xAI Grok at 34 million, and Anthropic Claude at 11 million. (The development comes as more Android users have been found to engage with Gemini directly through the operating system compared to using the standalone Gemini mobile app, suggesting it may have a broader user base. Last month, Google said the Gemini app had over 650M monthly active users.)
• The European Commission opens a probe into Meta after it changed its business API policy in October 2025 to ban general-purpose chatbots from the messaging app, stating the API isn't designed to be a platform for the distribution of chatbots. (The change goes into effect in January 2026. But the fact that Meta's own AI product Meta AI is accessible, while the others cannot has prompted concerns that the policy may "prevent third-party AI providers from offering their services through WhatsApp" in the bloc and that "competing AI providers may be blocked from reaching their customers through WhatsApp.")
• Swiss government urges people to ditch Microsoft 365 and others due to lack of proper encryption.
• Google gives Drive's file viewer an overhaul with improved support for third-party file formats, such as PDFs, videos, images and audio files, including a new toolbar and app bar and an option to search by transcript when looking at a video.
• Google adds the ability to auto-fill account information in Chrome, as well as travel details and loyalty cards saved in Google Wallet.
• Coinbase reopens app registrations in India after an over-two-year pause, letting users make crypto-to-crypto trades; plans to open up crypto buying in 2026.
• Taiwan bans Chinese social media app Xiaohongshu for one year, citing widespread scams and the company’s failure to comply with local regulations.
• Meta updates Facebook's Feed, search and navigation systems to help users "find what you are looking for and dive deeper into your interests"; says it's "streamlining how you create content in the app with a redesign of how you create Stories and Feed posts, as well as how you interact with and share them."
• Brave debuts a new agentic AI browsing feature that leverages Leo, its privacy-respecting AI assistant, to perform automated tasks for the user; says it runs on a separate, isolated profile to minimise the risk of prompt injection attacks and block access to the browser's settings page, non-HTTPS sites, the Chrome Web Store, and any sites flagged as suspicious.
• U.S. grocery delivery service Instacart gets caught using AI to charge wildly different prices for the same item without the stores' knowledge, according to a report from Groundwork Collaborative, Consumer Reports and More Perfect Union.