Tech Roundup: ChatGPT Health, Gmail AI Inbox & More

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

• Meta demoes SeamlessExpressive, an AI that translates peoples words into different languages and edits their mouth movements to match.
• Google updates its search bard on the home page with a new "+" at the far left that replaces the magnifying glass with option to upload an image or file.
• LG updates its smart TVs with support for Microsoft Copilot with no option to uninstall the app.
• Apple adds Google Cast support to its Apple TV app for Android, allowing users to stream content on their TV; comes as Disney+ becomes available to stream on Meta's quest headsets.
• Apple officially expands Fitness+ to 28 new markets, including Chile, Hong Kong, India, the Netherlands, Singapore and Taiwan, giving users access to 12 different workout types and dubbed versions of workouts and meditations in Spanish, German and Japanese.
• Google says it will now publish new source code to the Android Open Source Project twice a year, down from releasing source code for each quarterly Android release to ensure platform stability for the Android ecosystem and better align with Android's trunk-stable development model; says it helps simplify development, eliminates the complexity of managing multiple code branches and allows them to deliver more stable and secure code to Android platform developers.
• OpenAI adds Apple Music support and upgraded image generation capabilities to ChatGPT.
• Roblox says it's willing to make changes to have its ban lifted in Russia, which is one of its biggest markets, weeks after it was banned in the country for allegedly hosting extremist materials and LGBT propaganda.
• Austrian privacy non-profit noyb files complaints against TikTok, AppsFlyer and Grindr, accusing the popular video sharing platform of unlawfully tracking users across apps.
• Google debuts a new Feeds app for Google Chat, allowing users to news, blog posts and industry research from any Atom or RSS feed directly into their group conversations and spaces.
• Austria's high court orders Meta to change its personalised ad practices, ruling it as illegal and setting a legal precedent across the E.U.
• Google's AI experiment Doppl receives a major upgrade with a personalised Fashion Discovery Feed, built-in links to directly shop your looks and a Collections feature to save favourite fits.
• Google's YouTube tests image posts in the Shorts feed and hiding dislike button in Shorts within the three-dot overflow menu, in some cases renaming it to "Not Interested"; tests the ability to tweak the background of shared lyrics via YouTube Music using AI.
• Google debuts a new UPI-powered credit card service Flex within Google Pay in India; launches it in partnership with Axis Bank, with plans to add more issuer partners "very soon."
• Apple and Google allow developers of mobile applications to distribute their wares through third-party app stores and accept payments from alternative payment providers to comply with Japan's Mobile Software Competition Act (MSCA).
• Meta plans to integrate Singaporean startup K-ID's AgeKey technology into its apps beginning next year as a more user-friendly way to implement age checks. (Using the same standards-based technology as passkeys, AgeKey lets users verify their age once and reuse the same credentials across compatible apps, allowing for more uniformity for age verification. K-ID said AgeKey does not disclose a user's exact age, birth date, or other personally identifiable information, only indicating whether the user meets the required age criteria and how their age was verified.)
• Data from Cloudflare shows that iOS devices generated 35% of mobile device traffic globally in 2025, with internet traffic registering a 19% growth; Google, Facebook, Apple, Microsoft, Instagram, Amazon AWS, YouTube, TikTok, Amazon and WhatsApp take the top 10 spots for the most popular internet services worldwide.
• Alphabet's Waymo is reportedly working on a Google Gemini-powered in-car AI assistant, as the company, which paused its service during power blackouts in San Francisco on December 20, says it is updating its fleet to improve navigation during widespread outages.
• Google says it's gradually rolling out an option for users to change their Google gmail account address to a new email address that ends in gmail.com, with the old email address set as an alias. (It's worth noting that the process is equivalent to setting up a second account as emails will be sent to both old and new email addresses, and the old account will continue to exist.)
• China plans to restrict artificial intelligence-powered chatbots from influencing human emotions in ways that could lead to suicide or self-harm, according to new draft rules released by the Cyberspace Administration; requires minors to get guardian consent to use AI for emotional companionship, with time limits on usage.
• Meta agrees to acquire Singapore-based AI startup Manus (it was originally based in Beijing and Wuhan); says it will continue to operate and sell the Manus service, adding the company will cut Chinese investor ties after the deal and "will discontinue its services and operations in China." (The purchase, however, has come under scrutiny, with Chinese reviewing the deal for possible technology export control violations.)
• The U.S. state of New York passes law requiring social media companies to display warnings about the potential mental health harms of using their products; says "to combat the mental health risks of using harmful features of social media platforms that prolong use, this legislation will require social media companies to display warning labels on their platforms when a young user initially uses the predatory feature and periodically thereafter, based on continued use."
• More than 21% of YouTube videos accessible via the platform is "AI Slop" or brainrot content, according to a report from Kapwing.
• Google begins rolling out a redesigned Reading Mode for Chrome on Android with options to change font, size and background colour.
• Internal docs reveal Meta's tactics to fend off pressure to crack down on scammers, including efforts to make scam ads "not findable" for regulators and others.
• Pebble unveils Pebble Round 2 smartwatch with a rounded screen and basic health and activity tracking (no heart rate monitor) for $199.
• xAI launches Grok Business, priced at $30 per seat/month, for SMBs, and Grok Enterprise for larger organisations.
• A Guardian investigation finds that Google's AI Overviews -- the generative AI summaries that appear at the top of search results -- are serving up inaccurate health information, putting people at risk of harm; Google says many of the health examples shared with them were "incomplete screenshots" and that the vast majority of its AI Overviews were factual and helpful. (The development comes amid growing concerns about the reliability and trustworthiness of AI chatbot responses when it comes to news and financial advice.)
• London readies to become the first city where both U.S. and Chinese robotaxis operate, as Waymo and Baidu prepare to launch their robotaxis in the city in 2026.
• NVIDIA announces the Alpamayo family of AI models, tools, and datasets for autonomous vehicles; comes as Universal Music Group announces a partnership with NVIDIA "to pioneer responsible AI for music discovery, creation, and engagement," centered on NVIDIA's Music Flamingo model.
• Boston Dynamics partners with Google DeepMind to integrate Gemini Robotics models into Atlas robots, boosting their object-manipulation capabilities and more.
• Meta unveils teleprompter and EMG handwriting features for Meta Ray-Ban Display and expands Pedestrian Navigation for the glasses to four new U.S. cities, as it pauses the release of its Ray-Ban Display glasses to the U.K., France, Italy and Canada due to "unprecedented demand and limited inventory"; Meta and Garmin show an early demo of using Meta's wrist-based neural band inside of a car to control an infotainment system, as part of Garmin's Unified Cabin.
• Amazon's Ring revamps its home sensors, including door, window, and break glass sensors built on the Amazon Sidewalk protocol, and expands the protocol internationally.
• Meta's Threads pilots games in chats, as it continues to expand its Communities feature with more topics to attract more users to its platform.
• Google says it's working on major Snapseed update for Android with a complete redesign and a brand new icon, more than six months after it released the update for iOS in June 2025.
• China unveils rules banning major platforms like Alibaba from forcing merchants into discounts or practices seen as disrupting market order; to come into effect starting February 2026.
• Web browser extensions become a new way to funnel users' interactions with AI chatbots (amounting to hundreds of millions of prompts per month from millions of active users) to companies like Profound that sell market insights based on activity in answer engines like ChatGPT, Gemini and Claude without their explicit consent. (While Profound claims the data is "anonymised, aggregated, scrubbed of PII, and compliant with GDPR and CCPA," the reality is that browser extensions with the "Read and change all your data on all websites" permission can access chats from the web browsers, posing severe commercial and personal privacy risk. Once siphoned, they may be sold, packaged, and resold. The best practice is to only use web-based AI chats in Incognito/Private mode where the default is to not allow extensions to run.)
• A new report from Similarweb reveals that OpenAI ChatGPT is losing its market share to Google Gemini on the web, with its share eroding from 86.7% a year ago to 64.5% in January 2026, as Gemini surges from 5.7% to 21.5%.
• Chinese artificial intelligence firm DeepSeek agrees to launch the first country-specific version of its namesake chatbot for Italian users; also commits to "making its disclosures about the risk of hallucinations more transparent, intelligible and immediate" to meet the requirements of regulators.
• OpenAI debuts ChatGPT Health, which lets users import medical records and other data from health apps into ChatGPT; begins rolling out a HIPAA-compliant version of ChatGPT for clinicians to assist with medical reasoning and administrative tasks.
• Google unveils an AI Inbox view for Gmail, showing users to-dos and summaries of topics rather than a traditional email list; makes Help Me Write, AI email summaries and Suggested Replies free to all Gmail users.
• Microsoft debuts Copilot Checkout in the U.S., enabling in-chat purchases via PayPal, Shopify, Stripe and other merchants.
• Google's YouTube adds the option to filter content by "Shorts," as it removes the option to sort by Relevance, Upload date, View count, and Rating with a new "Prioritise" filter to sort by Relevance or Popularity.
• Internet connectivity in Iran collapses amid nationwide protests following a sharp drop in the value of the country's currency.
• Italy fines Cloudflare €14.2 million for refusing to block access to pirate sites on its 1.1.1.1 DNS service as part its controversial Piracy Shield law, which has attracted scrutiny for blocking hundreds of legitimate websites; Cloudflare calls the fine unjust and says it's considering discontinuing its services in the country.
• Anthropic cuts off xAI’s access to its models for coding amid heightened competition between top AI companies; adds safeguards to prevent third-party apps, like OpenCode, from spoofing Claude Code to access Claude models for more favourable pricing and limits.
• Planet and Google team up for a research effort that examines whether computing infrastructure typically housed on Earth can function in orbit.
• Amazon gets accused of listing products from independent retailers without their consent or knowledge as part of an experimental AI tool called Buy For Me that duplicates listings and handles purchases without leaving the service by turning into a search engine of sorts, even as the company has ironically sued Perplexity for scouring the web for products -- including Amazon -- to buy products.
• A new report from WIRED reveals that OpenAI is asking contractors to upload projects from past jobs to evaluate the performance of its AI agents, leaving it to them to strip out confidential and personally identifiable information.
• Instacart says it's "ending all item price tests" and "retailers will no longer be able to use Eversight technology to run item price tests on Instacart"; claims they are a form of short-term, randomized A/B testing, and that it's not doing surveillance pricing.