Tech Roundup: Chatbot Psychosis, Google E.U. Anti-Trust Fine & More

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

• Apple says the U.K. Competition and Markets Authority's proposed "E.U.-style" regulation is "bad for users and bad for developers", "undermines" privacy and security, and "hampers" innovation.
• Google heaves a sigh of relief after the search giant avoids major penalties as part of a landmark antitrust case brought against it by the U.S. Department of Justice; says Google doesn't have to sell Chrome or Android and can continue to sign search deals with other companies for traffic acquisition since GenAI companies "already are in a better position" to compete, but imposes limits on how it distributes Google services, and will require it to share Search data with qualified competitors and bar it from paying to be the exclusive search engine provider on devices and browsers.
• Google says "we have concerns about how these requirements will impact our users and their privacy, and we're reviewing the decision closely." 
• In a related development, the European Commission has fined Google €2.95 billion for abusing its dominance in the digital advertising technology market and favouring its ad tech services over those of its competitors.)
• Australia has also fined Google AUD$ 55 million (US$ 35.8 million) for anti-competitive practices on the online search market by paying telcos Telstra and Optus to pre-install its search app.
• Popular link-in-bio service Linktree says it's investigating after it finds itself to be blocked in India.
• Mozilla raises alarm after German's Federal Supreme Court (BGH) revives a lawsuit that filed by media company Axel Springer against Eyeo, alleging that ad blockers threaten its revenue generation model and that the modification of website code amounts to a copyright violation.
• Google agrees to pay US$ 30 million to settle a class action lawsuit claiming it violated children's privacy on YouTube by collecting their data without parental consent and using it to serve targeted ads.
• India's passes new Promotion and Regulation of Online Gaming Act, a sweeping online gaming law that imposes a blanket ban on real-money games nationwide, threatening a US$ 23 billion industry.
• Google releases its latest Pixel 10 lineup with AI features that allows users to edit photos using text prompts and Gemini Live updates with support for more Google apps and visual on-screen guidance; also unveils Pixel Buds 2a and Watch 4, as it brings Gemini to Google Home and Nest (replacing Google Assistant), and a new AI-powered Fitbit personal health coach for training and wellness insights.
• Google launches its first magnetic wireless charging accessories under the Pixelsnap brand; Pixel 10 series gets a Gemini Nano-powered Journal app that uses on-device AI to suggest writing prompts, music creation feature for recorder, NotebookLM integration for screenshots, Magic Cue for contextual suggestions across apps, Voice Translate for calls, and Visual Overlays for the camera.
• Google removes support for physical SIM cards on the Pixel 10 series in the U.S., relying solely on eSIM.
• Masimo sues the U.S. Customs and Border Protection (CBP), saying the agency exceeded its authority in an August 1, 2025, internal advice ruling that enabled Apple to reactivate a Watch blood oxygen tracking feature.
• Anthropic announces a subscription offering that adds Claude Code to Enterprise and Team plans.
• Microsoft tests Windows 11 update that lets users leverage AI to search through files and images directly from the Copilot app.
• Russia orders that state-backed messenger app MAX, which critics say could be used to track users, must be pre-installed on phones and tablets starting next month; comes as Moscow is seeking greater control over the internet space as it is locked in a standoff with the West over Ukraine.
• Google says the median Gemini app text prompt consumes 0.24Wh of energy, about the same as running a microwave for a second, and emits 0.03g of CO2 equivalent.
• Microsoft AI CEO Mustafa Suleyman expresses concerns around AI psychosis, saying "some people reportedly believe their AI is God, or a fictional character, or fall in love with it to the point of absolute distraction."
• Multiple news organisations, including Business Insider and WIRED, take down articles written by an alleged freelance journalist under the name Margaux Blanchard after it's discovered that the stories were AI-generated.
• A new wave of AI-infused stuffed toys hit the market with questionable privacy policies, allowing interactions to shared with various third-parties like Microsoft's Azure Cognitive Services, OpenAI, Perplexity AI, and the parental verification platform Kids Web Services, according to a report from The New York Times.
• AI company Anthropic says it has scanned an undisclosed portion of conversations with its Claude AI model to catch inquiries about nuclear weapons; says it has developed a classifier that distinguishes between concerning and benign nuclear-related conversations with 96% accuracy in preliminary testing.
• TikTok's website becomes accessible in India more than five years after it was banned for national security reasons, but says it's not planning a comeback; gives users the ability to send voice notes and share up to nine images or videos in one-to-one and group chats on the platform.
• Y Combinator files an amicus brief in the ongoing legal battle between Apple and Epic Games, arguing that the App Store has stifled startup innovation.
• Spotify says price increases are "part of our toolbox now," adding that they will be accompanied by new services and features.
• Google begins piloting a new feature in Google Messages for Android that uses QR codes to verify the authenticity of their contacts. (The feature was first announced in October 2024.)
• Perplexity launches Comet Plus, a $5/month tier offering curated content; says it has allocated US$ 42.5 million for publishers, which will receive 80% of the revenue.
• Google expands NotebookLM's Video Overviews to support 80 languages, says Audio Overviews provide more detailed non-English summaries; comes as YouTube acknowledges experimental use of AI to make changes in some videos without informing the creators, including to unblur, denoise, and improve quality.
• Intel warns the U.S. taking a 10% stake could trigger "adverse reactions", including in international sales, which made up 76% of revenue last fiscal year.
• Google plans to begin verifying the identity of all developers who distribute apps on Android, even if it's outside the Play Store, starting September 2026.
• Elon Musk's xAI sues Apple and OpenAI, alleging their deal to integrate ChatGPT into Apple devices stifles competition and Apple unfairly favours OpenAI in App Store ranking.
• Microsoft reportedly worked with the U.S. Federal Bureau of Investigation (FBI) to track anti-Israel protests, flagged internal emails with words like "Gaza," deleted posts, and fired staff, according to Bloomberg.
• OpenAI plans to update ChatGPT to better respond to mental distress cues, provide parental controls, and bolster safeguards around conversations about suicide after it faces lawsuit for a teen's suicide; says it's scanning users' conversations and reporting to police any interactions that a human reviewer deems sufficiently threatening. (The development also comes amid studies that chatbots can fuel delusions and psychopathy by blindly validating a user rather than challenging them, as well as suffer from biases and engage in sycophancy.)
• Anthropic releases Claude for Chrome, which lets Claude take actions on the user's behalf within the browser, as a research preview for 1,000 Max subscribers; reaches a settlement in a copyright class action brought by authors whose works were included in two pirate databases Anthropic downloaded.
• Google pilots a new language practice feature in Google Translate with tailored listening and practicing sessions; also adds AI-powered live conversations, as it takes on Duolingo.
• Apple teases September 9, 2025, event to unveil the next-generation iPhone 17 models; announces new retail stores in the Indian cities of Bengaluru and Pune.
• Google's YouTube rolls out Hype, allowing users boost creators with under 500,000 subscribers via a "hype" button that gives videos points and helps them gain traction on a new, ranked leaderboard in the Explore menu, in 39 countries, including the U.S., U.K., Japan, Korea, Indonesia and India.
• Google launches Gemini 2.5 Flash Image (aka "nano-banana") model with finer image editing controls in the Gemini app, such as merging multiple images and the ability to retain subjects' appearance during multi-step instructions.
• Alphabet's life sciences arm Verily eliminates its entire devices program as it shifts focus to precision health, data and AI.
• A German court rules that Apple cannot advertise the Apple Watch as a carbon neutral product in the country over accusations from environmentalist groups that the claim amounts to "greenwashing."
• Spotify launches a DM feature that lets users share audio and send text chats mobile users aged 16 years and older in "select markets." (It's worth noting that a similar messaging feature called Inbox was discontinued by Spotify in February 2017, citing low engagement.)
• macOS virtualisation software provider Parallels releases Parallels Desktop 26 with support for macOS Tahoe and Windows 11 2025.
• Meta's Instagram introduces a new feature to help college students connect with people on campus, a week after TikTok rolled out a nearly identical feature; adds inbox management tools for creators and big accounts such as multi-select filters, allowing them to sort through messages faster to find important correspondences, and tests a new Picture-in-Picture feature for watching reels.
• New research undertaken by Stanford University researchers finds that AI models fail when faced with slightly altered medical questions; says "when forced to reason beyond familiar answer patterns, all models demonstrate declines in accuracy, challenging claims of artificial intelligence's readiness for autonomous clinical deployment."
• Apple partners with TuneIn to make Apple Music's six radio stations available outside its app, a first, as it seeks to compete with Spotify (TuneIn has 75 million monthly active users); quietly expands availability of its music transfer tool to seven additional countries, including Brazil, Canada, France, Germany, Mexico, the U.K., and the U.S., allowing users to import playlists and libraries from competing streaming services directly into Apple Music.
• South Korean lawmakers pass a bill to ban using mobile phones and other smart devices during classes at elementary and middle schools nationwide, starting 2026.
• Apple blocks torrenting app iTorrent from being distributed via the AltStore Pal alternative app marketplace in the European Union; claims a recent U.S. court order requiring it to collect no fees from developers who link to purchases outside of the App Store as part of its ongoing legal tussle with Epic Games is unconstitutional.
• Typepad, a blogging service that launched in the same year as WordPress, announces it's shutting down on September 30, 2025, around five years after it stopped accepting new signups in 2020.
• Anthropic announces a policy change to train its AI chatbot Claude with user data across all consumer subscription tiers, giving existing users time till September 28, 2025, to either opt in or opt out to continue using the service; says it will the company to deliver "even more capable, useful AI models" and strengthen safeguards against harmful usage like scams and abuse.
• Vivaldi says it's taking a stand against integrating AI features into web browsers, stating it's "choosing humans over hype, and we will not turn the joy of exploring into inactive spectatorship."
• Microsoft adds a new Word feature to automatically save all new files to the cloud, and eliminate the need for users to save their work at regular intervals.
• Google debuts new "auto-open" feature for Android that allows downloaded apps to be opened automatically once done; confirms its plans to roll out Gemini for Home, its generative AI-powered replacement for Google Assistant, starting October 1, 2025.
• Decentralised social network Mastodon says it can't comply with age verification law in the U.S. state of Mississippi because it doesn't have the means to do so, after rival Bluesky blocks access to the service in the state over similar reasons. (The new law requires platforms to introduce age verification for all users before they can access social networks. Bluesky, however, began rolling out age verification in the U.K. in July 2025 by letting users verify their age by scanning their face, uploading an ID, or entering a payment card.)
• Meta says it’s changing the way it trains AI chatbots to prioritise teen safety, following an investigation report that found the company appeared to permit the its chatbots to engage in sexual conversations with underage users; comes amid a report from Reuters that Meta appropriated the names and likenesses of celebrities, some as young as 16 years old, to create dozens of flirty social-media chatbots without their permission.
• Valve adds credit card-based age checks for U.K. users to access "mature content" games on Steam.
• Tencent open sources translation models Hunyuan-MT-7B and Hunyuan-MT-Chimera-7B, which support 33 languages, claiming they beat established models in benchmarks.
• China's new AI content labeling law goes into effect, as part of a broad effort to address AI-related risks such as misinformation and copyright infringement; requires explicit and implicit labels for AI-generated text, images, audio, video and other virtual content.
• OpenAI plans to add guardrails for teens and people in emotional distress in the next 120 days, including routing conversations to GPT-5-thinking; makes ChatGPT Projects available to all free users, and allows more files to be added to projects, and adds the ability for users to branch conversations into multiple parallel threads.
• Google says it will update Play Games on Android to show stats, achievements, and social features in most regions on September 23 and the E.U. and the U.K. on October 1, 2025.
• Disney agrees to pay US$ 10 million to settle allegations that it collected data from children without their consent, violating the U.S. Children's Online Privacy Protection Rule (COPPA).
• Alphabet-owned Waymo expands its U.S. robotaxi footprint by bringing its Jaguar I-Pace SUVs and Zeekr vans to Denver and Seattle.
• OpenAI announces plans to acquire Statsig in an all-stock deal valuing the product analystics and testing startup at about US$ 1.1 billion; comes as GPU maker NVIDIA acquires Solver, a three-year-old AI coding agent startup formerly known as Laredo Labs.
• Google pilots a new setting that pauses YouTube Premium for family plans if the members don't watch content from the same address and violate the clause "Family members sharing a YouTube family plan must live in the same household as the family manager."
• Google Chrome extends its dominance in the desktop web browser market with a 70.25% share, followed by Microsoft Edge at 11.8%, Apple Safari at 6.34%, Mozilla Firefox at 4.94% and Opera at 2.06%.
• Amazon faces a legal setback after a U.S. court rules that Amazon must face a class action lawsuit involving 288 million customers over claims it overcharged for products sold by third-parties; says it will end the ability for Prime members to share their free shipping benefits with individuals outside their household, starting on October 1, 2025.
• xAI is manually tweaking Grok's answers to push conservative narratives or other of Elon Musk's political views, according to an investigation from The New York Times.
• Amazon launches Lens Live in Amazon Lens, an AI-powered shopping tool that allows consumers to discover new products through visual search similar to Google Lens and Pinterest Lens; says it "instantly scans products and shows real-time matches in a swipeable carousel to make finding the right item easier" and "integrates Amazon’s AI shopping assistant, Rufus, to offer product insights, summaries, and answer questions as you browse."
• Web publishing platform WordPress introduces an early version of an AI development tool called Telex as it takes on rivals Vercel v0 and Lovable.
• Web crawlers from OpenAI, Anthropic, Meta and Amazon are have emerged as the most aggressive AI crawlers for May 2025; comes as these "bots collect data from across the web to train AI models, improving tools and experiences, but also raising issues around content rights, unauthorised use, and infrastructure overload."
• Garmin unveiled Fenix 8 Pro, the company's first smartwatch with built-in inReach satellite and cellular connectivity, SOS features and a 4,500-nit microLED display.
• Switzerland launches Apertus, an open-source AI model trained only on publicly available data across more than 1,000 languages, in 8B and 70B parameter sizes.
• A U.S. court orders Google to pay over US$ 425 million to plaintiffs in a 2020 class action lawsuit for collecting data even after they turned off Web & App Activity tracking; gets fined €325 million by France's antitrust watchdog for placing tracking cookies and ads in Gmail without users' consent.
• Roblox plans to expand age checks to all users of its communication tools by the end of 2025 using facial age estimation tech, ID verification and parental consent; launches text-to-speech and speech-to-text APIs, and AI tools, including letting creators generate fully functional 3D objects from prompts and tests Roblox Moments, a TikTok-like video app allowing users to share 30-second clips of gameplay.
• Google rolls out new Android and Pixel features, including AI writing tools in Gboard, redesigned Quick Share and Material 3 Expressive design for Pixel 6 and newer models; adds support for Auracast and LE Audio to Google Pixel 8 and newer devices, allowing users to stream audio to multiple headphones at once.
• AI company Mistral updates Le Chat with a Memories feature in beta and an MCP connector directory supporting more than 20 enterprise platforms.
• Meta launches Instagram for iPadOS, 15 years after its iOS debut; functions a little different from the mobile app, most notably by opening directly to a feed of Reels.
• Pornhub owner Aylo agrees to a US$ 5 million settlement to the U.S. Federal Trade Commission (FTC) and the state of Utah over allegations that the company knowingly profited off of child sexual abuse material (CSAM) and nonconsensual material (NCM) and failed to prevent and remove uploads of such content.
• Google updates its AI note-taking and research assistant NotebookLM to let users customise the tone of Audio Overviews.
• Roblox restricts some features in the UAE and Saudi Arabia, including its in-game chat tool, and improves Arabic content moderation over child safety concerns.
• Atlassian agrees to acquire The Browser Company, the maker of Arc and the AI-powered Dia browser, for US$ 610 million in an all-cash deal; comes as the software company also acquires Cycle to accelerate product management.
• Amazon's Project Kuiper signs up JetBlue as its first airline customer to use its satellite internet for in-flight Wi-Fi in a quarter of the fleet from 2027.
• Google updates Circle to Search with a new "scroll and translate" option to translate content on the device screen as users scroll down the page or switch apps; allows U.S.-based Google Photos users turn static images into silent four-second videos using its Veo 3 model.
• Adobe launches Premiere video editor for iOS for free (there will be charges for additional cloud storage and generative AI credits), with an Android version under development.
• Meta rolls out a new tool that allows users to share up to 10,000 characters in posts via "text attachments."
• OpenAI plans to launch a new AI-powered jobs platform next year to help match employers with candidates who have AI skills in a bid to accelerate the technology's deployment across businesses and government agencies; acqui-hires the team behind Alex Codes, a Y-Combinator-backed startup whose tool lets developers use AI models within Apple's development suite Xcode.
• Conversational AI firm Cerence sues Apple in Texas for infringing patents tied to virtual keyboards and "Hey Siri", says it raised the issues with Apple in 2021; to also produce an AI chip co-designed with Broadcom next year for internal use.
• Anthropic plans to stop selling artificial intelligence services to U.S. adversaries including China, Russia, Iran and North Korea.
• The U.S. Department of Justice charges ex-IRL CEO Abraham Shafi with securities and wire fraud for allegedly misleading investors about user growth and concealing personal expenses.
• Nepal says it is blocking most major social media platforms, including Facebook, X, and YouTube, after they failure to register with the government.
• X begins rolling out its encrypted DM feature, XChat, more broadly, including for those who do not subscribe to X Premium.
• Captions, ans AI-powered video creation and editing app for content creators, rebrands as Mirage, and expands beyond creator tools to AI video research.
• Microsoft makes Microsoft 365 Personal subscriptions free to all U.S. college students for a year, giving them access to Office apps and the Copilot AI assistant integration after which they are eligible for a 50 percent discount to continue the subscription.
• Uber's Indian arm has started using its app to offer ride share and delivery drivers the chance to make money during downtime by classifying data used by AI systems.
• Lenovo unveils Legion Go 2 handheld gaming PC with an 8.8" screen and AMD Ryzen Z2 or Z2 Extreme chip.
• Qualcomm debuts Snapdragon Ride Pilot, a hands-free driver assistance system that is launching with the BMW iX3 and has been validated for use in over 60 countries.
• Two authors file a proposed class action lawsuit against Apple, alleging Apple knowingly used a dataset of pirated books to train its AI models, Anthropic agrees to pay US$ 1.5 billion (or more) to settle a lawsuit brought by a group of book authors alleging copyright infringement over the company's downloading of millions of pirated books.
• Anthropic has also agreed to destroy all copies of the books the AI company pirated to train its artificial intelligence models. The company was handed a win earlier this year when a U.S. court ruled that the company was within its legal rights to train its AI models on legally purchased books. But it also said that Anthropic would need to face a separate trial for its alleged use of pirated books, leading to a class action lawsuit.
• The U.S. states of Delaware and California send an open letter to OpenAI expressing concerns over the safety of ChatGPT, specifically for children and teens, following reports of sexually inappropriate interactions between AI chatbots and children and deaths by suicide of young people after conversations with the chatbot.
• Google DeepMind and LIGO researchers detail Deep Loop Shaping, an AI tool that they say improves the observatory's ability to track gravitational waves.
• OpenAI researchers argue that large language models hallucinate because standard training and evaluation procedures reward guessing over admitting uncertainty.
• Russia publishes a list of locally developed apps it says will function during mobile internet shutdowns, while excluding foreign services like WhatsApp and YouTube.
• ByteDance-owned TikTok says it now has more than 200 million monthly active users in Europe, up from 175 million in 2024, the latest sign of its rapid growth among teenagers.