Tech Roundup: Meta AI Studio, Twitch Revamp & More

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

• The U.S. Justice Department sues TikTok, saying it collected children's data without parents' permission and knowingly let users under the age of 13 create accounts.
• Turkey blocks Instagram within the country for failing to comply with "laws and rules"; reports say the decision is motivated by the platform's blocking of condolence posts following the assassination of Hamas leading Ismail Haniyeh.
• Privacy-focused search engine DuckDuckGo gets blocked in Indonesia by its government following complaints about pornographic and online gambling content in its search results.
• China announces proposals for a national internet ID to protect privacy and prevent online fraud, a move that critics say could further concentrate government monitoring and lead to social control.
• The European Union's AI Act takes effect, triggering staggered compliance deadlines for various AI developers and apps.
• The U.S. Justice Department defends new law requiring the sale or ban of TikTok in the country, stating the platform collected data about its users' views on gun control, abortion, and religion and censored content at ByteDance's direction.
• Meta releases Segment Anything Model 2 with support for object segmentation in videos and images and claims it's "the first unified model that can identify which pixels belong to a target object in an image or video"; rolls out AI Studio in the U.S., letting users create and share AI chatbots based on their interests, including enabling Instagram creators to set up chatbots to answer common DM questions and story replies. (The debut of AI Studio has had the side effect of Meta discontinuing celebrity AI chatbot feature less than a year after launch.)
• The European Commission unconditionally approves HPE's US$ 14 billion takeover of networking gear maker Juniper Networks after finding no competition concerns.
• Twitch rolls out a revamped iOS and Android app, prominently showing a TikTok-like feed of recommended content; comes amid reports that the Amazon-owned livestreaming site generated about US$ 667 million in advertising revenue and US$ 1.3 billion in commerce revenue in 2023, making up less than 0.5% of the e-commerce giant's overall revenue after its acquisition a decade ago.
• Apple debuts Apple Intelligence in the iOS 18.1, iPadOS 18.1, and macOS 15.1 developer betas; says its AI models were trained on Google's custom chips and that it no private user data was included.
• LinkedIn, Pinterest and TikTok are testing programs that let publishers sell ads on their videos and pages, respectively, in exchange for a cut of the resulting revenue, according to a new report from The Information.
• A U.S. court in New York rules that border agents must get a warrant before searching the electronic devices of U.S. and international travellers crossing the border.
• Google updates its Files app for Android with a new Recents panel design that presents files in a carousel format with previews; tests a redesign of the video editing user interface in Google Photos with options to adjust video playback speed.
• Ride hailing service Uber debuts new feature in India that allows user to book as many as three concurrent rides, as it takes on local competitor Ola that supports two concurrent rides.
• Pinterest reports 522 million monthly active users in Q2 2024, as internal figures show X has 250.8 million daily active users in Q2, up 1.56% YoY; Snap reports 432 million daily active users for the same period, up 9% YoY.
• OpenAI begins rolling out ChatGPT's Advanced Voice Mode to ChatGPT Plus users; showcases a new safety technique built into GPT-4o Mini called instruction hierarchy, which is a way to prioritise the developer's original prompts and instructions over any potentially manipulative or malicious user-created prompts. (The development comes as the company said it's "dedicated" to "rigorous safety protocols at every stage of our process" and is working with the U.S. AI Safety Institute to provide early access to its next major generative AI model for safety testing.)
• The U.S. Consumer Product Safety Commission (CPSC) rules that Amazon is legally responsible for recalled products sold on Amazon.com, even by third-parties, and must remove them and notify purchasers.
• The U.S. Senate passes the Kids Online Safety Act and the Children and Teens' Online Privacy Protection Act (aka COPPA 2.0) to safeguard children from online harms and ban targeted advertising to those under 17; U.S. senators also unveil the bipartisan NO FAKES Act, which would make individuals and companies liable for producing, hosting, or sharing unauthorised AI deepfakes.
• Spain's antitrust watchdog fines Booking.com €413.2 million for abusing its dominant position of 70%-90% market share in the country over the last five years following allegations that the company bans hotels from offering lower prices on their own sites than on Booking.com, while imposing price discounts on hotel rooms without consulting the hotels; says "these practices have affected hotels located in Spain and other online travel agencies that compete with the platform."
• Russian lawmakers pass a bill to allow businesses to use crypto currencies in international trade, as part of efforts to skirt Western sanctions imposed after Russia's invasion of Ukraine.
• Meta agrees to pay US$ 1.4 billion to settle a lawsuit filed by the state of Texas accusing the company of using facial recognition tech to collect biometric data of millions of Texans without their permission. (The settlement concludes a lawsuit brought by the state in February 2022, which alleged that Meta unlawfully captured Texans' biometric data without obtaining their informed consent. It's worth noting that Meta agreed to pay a US$ 650 million settlement in a class-action lawsuit in Illinois under a state privacy law over similar allegations related to its face-tagging system in 2021.)
• Microsoft calls on the U.S. Congress to pass a comprehensive law to crack down on AI deepfakes that aim to interfere in elections or maliciously target people; removes ads from Skype for a "smoother, decluttered and more enjoyable user experience."
• AI company Perplexity launches a program to share ad revenue with partners such as Time, Der Spiegel, Fortune and WordPress.com, after weeks of plagiarism accusations.
• The U.K. Competition and Markets Authority (CMA) begins preliminary investigation into whether Google's partnership with Anthropic qualifies as a merger and harms competition in the country; comes as Microsoft accuses Amazon and Google of "muddying the waters" in connection with CMA's ongoing probe into its licensing practices.
• Canva announces plans to acquire Leonardo.ai, an Australian generative AI content and research startup, as part of its goal to build a "world-class suite of visual AI tools."
• Chipmaker Qualcomm debuts the 4nm Snapdragon 4s Gen 2 Mobile Platform, a feature-limited version of the Snapdragon 4 Gen 2, to make 5G accessible to smartphones in the sub-$100 category.
• Walmart-backed Flipkart consolidates its various payment and fintech offerings, including UPI and BNPL, under Flipkart Pay.
• Apple expands Emergency SOS via satellite to Japan with the release of iOS 17.6 update; files a motion to dismiss the U.S. Justice Department's antitrust lawsuit, arguing the complaint has not demonstrated anticompetitive conduct or that it is a monopoly power in the U.S. smartphone market, and that forcing it build tools for competitors could have a "chilling" effect on innovation.
• Music streamer Spotify ups the limit on the number of lyrics free users can view after putting the feature behind a paywall back in May 2024.
• Meta says its family daily active people is up 7% to 3.27 billion on average for June 2024; updates Threads to simplify exploring trending topics on the platform with a new "blue label above posts that are related to a recent trending topic" as it hits 200 million monthly users.
• Milan-based mobile app developer Bending Spoons acquires Dutch file-sharing service WeTransfer for an undisclosed sum, after purchasing Evernote in November 2022.
• Amazon sues Nokia in a U.S. federal court, accusing the Finnish telecom company of infringing 12 AWS cloud computing patents.
• E-commerce platform Etsy announces plans to begin testing its first loyalty program in September to boost sales.
• Google rolls out online safety features to make removing explicit deepfakes from Search at scale easier and to prevent them from appearing high up in results; makes it possible to use its "About this image" tool with Circle to Search or Google Lens as well as improves "About this result" on Search to get "more context on the online sources you see."
• Google updates Maps with options to easily report incidents; adds camera alerts in Waze that are used to enforce things like speed limits or carpool lanes and the ability to share real-time traffic information with friends.
• Nothing unveils the Phone 2A Plus, offering the same 6.7-inch display as the 2A and a new MediaTek Dimensity 7350 Pro 5G SoC, for US$ 399.
• Google debuts new Gemma 2 2B billion featuring built-in safety advancements and a powerful balance of performance and efficiency, alongside ShieldGemma classifier models to detect and mitigate harmful content in AI models inputs and outputs such as hate speech, harassment, sexually explicit content and dangerous content.
• Microsoft adds strategic partner OpenAI to its list of competitors, alongside Amazon, Apple, Google and Meta, days after the AI upstart announced a prototype of a search engine called SearchGPT.
• Apple reports US$ 24.21 billion in revenue from its Services division, which includes App Store, Apple TV+s and Apple Music, in Q3 2024, creating an all-time record, as iPad sales soar amid a slump in iPhone revenues; Amazon reports ad sales of US$ 12.77 billion for Q2 2024, up 20% YoY.
• Telegram rolls out a Mini App Store to help users find apps, a new in-app browser that supports multiple tabs; says mini apps have more than 500 million monthly users.
• Google unveils Gemini-powered AI features for Chrome on desktop, including support for Google Lens, comparing products across tabs using Tab compare when shopping online, and searching history via natural language queries; expands School time, a parental control feature, to more devices and begins rolling out cross-device services to devices running Android 11 and up.
• Music AI startup Suno, which allows users to create songs from text-based prompts, acknowledges its product was trained "essentially all music files of reasonable quality that are accessible on the open internet," citing a Fair Use argument to say that the law should allow for AI training on copyrighted works without permission or compensation; comes after Udio and Suno were sued in June by music labels over allegations that they trained their AI models by scraping copyrighted materials from the internet.
• The development has also triggered huge increase in the number of websites that are trying to block crawlers from companies in the wake of the generative AI boom. While the rush for data needed to train AI models has been deemed unethical, it also highlights how companies that have enormous amounts of data are now in a position where they have the power lock up data that was previously available to anyone, upending decades of norms across the internet.
• Google signs a non-exclusive agreement with Character.AI to use its technology after it acquihires CEO Noam Shazeer and co-founder Daniel De Freitas in a surprise move.
• Apple reportedly puts pressure on Chinese tech giants Tencent and ByteDance to close loopholes used by their in-app creators to funnel users to external payment systems and circumvent its typical 30% commission on purchases made from within the apps.
• Newly formed German AI startup Black Forest Labs, founded by researchers who developed the technology behind Stable Diffusion and invented the latent diffusion technique, announces the release of its first suite of text-to-image AI models, called FLUX.1.