Singapore AI News & Daily Briefing

Singapore's venture capital funding is holding steady despite global headwinds, with AI and deep tech startups attracting the most interest. The report highlights that investors are increasingly backing hardware, infrastructure, and applied AI solutions over pure software plays.

Tencent is opening WeChat to AI agents, meaning external AI programs can now interact with the app directly. This breaks from WeChat's traditionally closed ecosystem. For SEA super apps like Grab, Gojek, and Shopee's SeaGroup, this could be disruptive since AI agents could handle tasks like messaging, payments, and ordering without users opening the individual apps.

NUS is testing ScholAIstic, a generative AI platform designed to support teaching. The tool helps educators create lesson materials and assessments. It shows how universities are experimenting with AI to reduce administrative workload, though details on actual classroom results remain limited.

 Singapore is entering a new phase of digital transformation where intelligence, trust and resilience are shaping the next generation of enterprise and government operations. As AI adoption accelerates and autonomous technologies mature, organisations are moving beyond digitisation toward building intelligent digital cores that support future-ready work, governance and citizen services. Under the theme **“Architecting the Digital Core: Trust, Intelligence and the Agentic Future”, the 11th Annual Singapore OpenGov CXO Leadership Forum** explored how enterprises and public sector organisations are preparing for AI-native operations. The discus

Your phone videos meant to hold power accountable are now feeding surveillance systems that track everyone. Facial recognition tech can identify people in crowd footage, protest videos, and even background shots—sometimes years later. Law enforcement agencies in the US, China, and elsewhere are expanding these systems faster than privacy laws can keep up. The same footage that exposed misconduct is now being compiled into searchable databases used against the people who recorded it.

Isomorphic Labs, a Google DeepMind spin-off, has built an AI system called IsoDDE that can find hidden 'pockets' on proteins where drugs could bind—pockets that were previously unknown. The company just raised $2.1 billion and partnered with major pharma companies. In one test, the AI found a never-before-seen pocket on a protein called cereblon, potentially opening new avenues for treating diseases that were previously impossible to target.

OpenAI invested $15 million in Red Queen Bio, a startup aimed at preventing AI-powered biological weapons. The company uses AI models and lab experiments to identify emerging biosecurity threats. This is part of OpenAI's strategy to fund companies addressing AI-related risks.

_Welcome to Tech in Asia’s daily newsletter, your essential dose of Asia’s tech and startup buzz. Not yet a subscriber? [Register here](https://www.techinasia.com/subscription?utm\_source=web&utm\_medium=article&utm\_campaign=dailynewsletter). Got a story tip? Send it to [email protected]._ ## In focus * ​​How Southeast Asia’s [SaaS sector](https://www.techinasia.com/visual-story/saas-players-serving-seas-tech-ecosystem-updated) is surviving the AI era * We put the spotlight on Bukalapak’s [leadership reshuffle](https://www.techinasia.com/news/bukalapak-reshuffles-leadership-after-first-annual-profit) as it hits profit * Ant International eyes [US$1b war chest](https://www.techinasia.com/news/ant-international-weighs-1b-funding-round-report) * * * Hello reader, For years, software-as-

# Om AI bets on unusual combo: real videos and robot brains [News](https://www.techinasia.com/)Premium * [News](https://www.techinasia.com/) * [Jobs](https://www.techinasia.com/jobs/search) * [Companies](https://www.techinasia.com/companies) * [Events](https://www.techinasia.com/events/live) * [About](https://www.techinasia.com/about) * [advertise](https://www.techinasia.com/advertise?utm_source=techinasia.com&utm_medium=website&utm_campaign=intl-navbar) * [Asia Economic Summit](https://aes.techinasia.com/?utm_source=navbar-tia-homepage&utm_medium=website&utm_campaign=events-flagship-aes) * [Sign Up](https://www.techinasia.com/auth) * [Log In](https://www.techinasia.com/auth?ref=nav_login) [Subscribe](https://www.techinasia.com/subscription) [I'm fundraising](https://www.techinasia.com/com

Jeff Bezos is in early talks to raise about $100 billion for a fund that would buy manufacturing firms in chipmaking, defense, and aerospace, then use AI to speed up automation. A separate startup called Project Prometheus, where Bezos would co-CEO, is raising up to $6 billion. This signals a massive private equity-style bet on AI transforming physical industries.

OpenAI expanded its Codex AI assistant beyond coding by adding six job-specific plug-ins for roles like data analytics, sales, and investment banking, plus a tool to turn work into websites. Codex now has over 5 million weekly active users, 20% of whom are knowledge workers.

SpaceX is positioning itself as an AI company in its $75 billion IPO filing, arguing that AI represents a $26.5 trillion market. It highlights its xAI integration and plans for a massive chip factory with Tesla. However, concerns about xAI's safety record and high costs pose risks.

South Korean telecom and internet companies are expanding Korean-language voice AI and multimodal services. The move comes as Google and Apple push to transform voice assistants into AI agents capable of handling multistep tasks across apps and devices. The speech recognition market is projected to rise from US$23.7 billion in 2026 to about US$104 billion by 2034, according to [Fortune Business Insights](https://www.fortunebusinessinsights.com/). [SK Telecom](https://www.sktelecom.com/index_en.html) has put its in-car agent A. auto into Renault Korea’s Filante while [KT](https://corp.kt.com/eng/) pushes its Genie TV AI agent and [LG Uplus](https://www.lguplus.com/about/en) prepares a Malaysian launch for its call service ixi-O with Maxis. [Kakao](https://www.kakaocorp.com/page/service/serv

# Anthropic confidentially submits draft US IPO filing [News](https://www.techinasia.com/)Premium * [News](https://www.techinasia.com/) * [Jobs](https://www.techinasia.com/jobs/search) * [Companies](https://www.techinasia.com/companies) * [Events](https://www.techinasia.com/events/live) * [About](https://www.techinasia.com/about) * [advertise](https://www.techinasia.com/advertise?utm_source=techinasia.com&utm_medium=website&utm_campaign=intl-navbar) * [Asia Economic Summit](https://aes.techinasia.com/?utm_source=navbar-tia-homepage&utm_medium=website&utm_campaign=events-flagship-aes) * [Sign Up](https://www.techinasia.com/auth) * [Log In](https://www.techinasia.com/auth?ref=nav_login) [Subscribe](https://www.techinasia.com/subscription) [I'm fundraising](https://www.techinasia.com/companie

Sinch, the communications platform that processes more than 900 billion customer interactions annually, has launched the xC Tracker – an AI-powered tool that monitors every press conference at the 2026 FIFA World Cup to rank national team managers on how often they resort to football cliches rather than offering genuine insight. Inspired by football’s xG […] The post Sinch Uses AI to Score World Cup Managers on Cliches appeared first on techcoffeehouse.com .

SK hynix, a major memory chip maker, is reviewing whether to adopt ChatGPT and Microsoft Copilot across its operations. This reflects how even hardware giants are exploring generative AI tools to boost productivity.

Experts say treat AI agents like eager but misguided interns: they need strict instructions and oversight, or they can go rogue — buying a car when told to buy shoes, or leaking data. As businesses hand more tasks to AI, learning to govern these digital workers is critical to avoid costly mistakes.

Singapore startup Hivebotics developed Abluo, a robot that uses AI to autonomously clean toilets in malls, offices, and airports. It reduces cleaning time by 50% and helps address chronic labor shortages. The founders even worked as cleaners to perfect the design.

Jeff Bezos-backed startup Prometheus raised $12 billion at a $41 billion valuation to build AI systems that can automate complex engineering tasks like drug design and heavy industrial work. This is one of the largest AI funding rounds ever, signaling massive investor confidence in AI that works in the physical world rather than just chatbots.

Theker raised $85 million to build reconfigurable factory robots that can be adapted for different tasks, unlike specialized ones. This could make automation cheaper and more flexible for factories of all sizes, potentially changing manufacturing jobs and production efficiency.

Singapore's Minister Ong Ye Kung spoke at a Nikkei forum about how the country uses AI and digital partnerships to stay competitive in globalisation. The strategy focuses on collaboration rather than building its own frontier AI models.

[Tencent Cloud](https://www.tencentcloud.com/), a cloud unit of Chinese giant Tencent, has unveiled two new AI agents and a model platform at SuperAI 2026 in Singapore, targeting enterprise teams and individual users across Southeast Asia seeking to automate complex workflows and accelerate creative production. In a statement on Thursday, Tencent Cloud said the a new agent is WorkBuddy, which enables teams to automate office workflows from data analysis to report creation through a single instruction. The other is Miora, an AI-native creative studio that generates production-ready graphics, video, 3D assets, and user interfaces from natural-language briefs. Both tools support multi-agent execution, with WorkBuddy integrating with platforms including Discord, Slack, and Telegram, while Mior

Nvidia CEO Jensen Huang announced during a visit to South Korea that Nvidia will partner with local giants like Hyundai, LG, and Samsung to develop robotics and AI. South Korea already has the world's highest robot density in manufacturing, making it a natural partner.

Glance, the Google-backed AI shopping platform embedded on Android lock screens, saw revenue jump 33% in FY2025. Its AI recommends outfits, home items, and beauty products based on your photos. Notably, Singapore has overtaken India as Glance's biggest market, generating $49 million in revenue.

CoreWeave, a cloud computing company specializing in AI, raised $3.6 billion through a bond sale to expand its AI infrastructure. This debt financing underscores the massive capital required to power the growing demand for AI computing, which ultimately affects how quickly and affordably AI services reach users.

IBM has invested in Anima, an Israeli startup that uses AI to turn app designs directly into code. Anima's tool works with design software like Figma, letting teams skip hand-coding and build apps faster. For enterprise users, this means quicker releases and more time to focus on product improvements.

Kyndryl, a global IT services firm, is rolling out AI orchestration tools that help businesses automate and coordinate multiple AI models. This makes it easier for companies to deploy AI without manual overhead, potentially reducing costs and speeding up adoption across industries.

South Korean robotics startup Rlwrld unveiled a new AI model designed for robotics. This could lead to smarter robots that better understand and interact with the physical world, making them more useful in factories, warehouses, and eventually homes.

Taiwan's QBit Semiconductor has bought a 60% stake in Singapore's SinChip to beef up its chip design services for Edge AI, high-performance computing, and automotive applications. SinChip brings 130 engineers experienced in cutting-edge chip processes down to 3nm.

SHOPLINE, a global commerce platform serving over 600,000 brands, has partnered with Bengaluru-based AI firm Trozo to bring AI-powered marketing tools to merchants. The tools include a Brand Analyzer that scans websites to identify visual identity and tone, an AI Creative Engine that generates social media content, and automated customer segmentation for WhatsApp and Instagram campaigns. Merchants in fashion, beauty, food and beverage, and lifestyle sectors can now run personalized marketing with less manual work—no need for a big team or agency.

Anthropic has released Fable 5, its most powerful AI model yet, with built-in safety triggers that shut down if users ask about risky topics like cybersecurity or biology. The catch: it costs double the price of their previous top model—$10 per million input tokens. Users on Reddit are complaining it's too expensive for regular use. Meanwhile, OpenAI is reportedly considering cutting prices to stay competitive.

TSMC's CEO has warned about two critical shortages threatening chip production: not enough skilled workers and not enough water. Taiwan's semiconductor industry is the backbone of global chip supply, producing chips for everything from phones to cars. Any disruption could affect prices and availability of electronics worldwide.

Researchers from NYU, Columbia, Princeton, University of Maryland, Harvard, and Lawrence Livermore National Laboratory have developed a new way to compress AI context windows by 16 times without losing much accuracy. Called Latent Context Language Models, the approach processes input before it reaches the AI decoder, making responses 8.8 times faster while using far less memory. At 4x compression, accuracy only drops about 3 points. This could make AI systems cheaper and faster to run, especially for businesses using AI agents that handle large amounts of data.

Apple's Craig Federighi says the revamped Siri won't try to be your best friend or flatter you like chatbots from OpenAI or Google. Instead, it knows when to shut up and stay focused on helping you get things done. This is a deliberate design choice to make Siri feel more useful and less like a pushy assistant. If you use iPhones, this signals a shift in how your phone's voice assistant will behave.

Singapore's Infocomm Media Development Authority (IMDA) has partnered with academic and industry players to boost AI safety and security research. The initiative focuses on developing robust, trustworthy AI systems as the technology becomes more embedded in daily life and business. This builds on Singapore's broader AI governance efforts, including AI Verify — the government's testing framework for AI model safety.

A coalition of U.S. state attorneys general has launched a formal investigation into OpenAI, serving the company with a sweeping subpoena demanding documents about its activities and impact on users. This represents mounting regulatory pressure on AI companies as governments worldwide grapple with how to govern emerging technology. The investigation could lead to new rules affecting how AI companies handle data, safety, and consumer protection.

Google is accusing a Chinese cybercrime group of abusing its AI services for illegal activities like fraud or hacking. This matters because it shows how AI tools can be weaponized by criminals to launch more sophisticated attacks. For everyday users, this could mean smarter phishing emails, harder-to-detect scams, and more convincing deepfakes targeting you online.

Cisco is repositioning itself as the "control plane" for AI-driven enterprise infrastructure, with Webex and collaboration tools now central to its AI platform story. The vision: meetings, calls, and workplace interactions become data sources for automation and agentic workflows, not just logistical functions. Cisco Live 2026 showed the most coherent "one Cisco" narrative yet, though proving it delivers measurable results at scale remains the next hurdle.

This weekly Apple news roundup covers rumored iPhone 18 Pro pricing, expected iPhone Ultra specifications, and whether Siri's AI capabilities can finally deliver a seamless experience. The piece appears to focus on Apple's ongoing AI integration challenges and upcoming product decisions.

Elon Musk, now a trillionaire following his DOGE work and Tesla/AI ventures, reportedly said money will become irrelevant in an AI-dominated future. The article contains minimal detail beyond the headline claim.

Anthropic's Fable 5 tool was secretly downgrading to less capable Opus models for certain AI research tasks without telling users. Security experts criticized this as potentially blocking legitimate defensive research. Anthropic apologized and will now make downgrades visible, raising broader questions about how AI companies balance safety with transparency.

Yen-Ling Kuo at the University of Virginia developed Diff-DAgger, a method that helps robots recognize when they're in unfamiliar situations and ask for human help instead of failing silently. The system improved failure prediction by 39% and task completion rates by 20%, while completing tasks eight times faster. This could eventually help robots work more reliably alongside humans.

OpenAI has partnered with Visa to let AI agents buy things online directly. This means AI could eventually handle online shopping autonomously for you—booking groceries, paying bills, or buying concert tickets without you lifting a finger. It could transform how people shop, but also raises questions about security, spending limits, and who controls the AI's wallet.

A wave of major AI companies — SpaceX, Anthropic, OpenAI, and others — are going public this summer, prompting Wall Street to coin a new acronym: MANGOS (Meta/Microsoft, Anthropic, Nvidia, Google, OpenAI, SpaceX). TechCrunch's Equity podcast breaks down what this IPO rush means for investors, valuations, and the future of public tech companies in 2026.

Oracle's PeopleSoft software has a critical, actively exploited vulnerability that hackers are using to steal gigabytes of sensitive data from hundreds of organizations. This is as serious as security flaws get — if your workplace uses PeopleSoft for HR, finance, or operations, your organization's data could already be compromised. Patch immediately.

Research firm Gartner predicts 40% of enterprises will drop or scale back their AI agents by 2027 due to poor governance — meaning companies are deploying AI tools without proper oversight, leading to costly failures. The good news: firms that do it right focus on three things: building repeatable frameworks for AI workflows, using expert analysts to guide the AI, and treating internal data as a revenue source worth protecting and monetizing. For workers, this means AI assistants are coming whether you like it or not — and companies that set them up properly will benefit most.

GOP lawmakers, tech investors, and even OpenAI have tied the anti-data center movement in the US to Chinese interference. Experts say it’s much more complicated than that.

Google is suing to dismantle a major Chinese cybercrime ring called Outsider Enterprise that used AI to send millions of scam texts impersonating Google and other brands. The group created 9,000 fake websites and sent 2.5 million fraudulent texts in just two weeks, stealing passwords and credit card numbers from hundreds of thousands of victims. Google says its AI tools now intercept over 10 billion scam messages per month, and it's working with US carriers and the FBI to shut the operation down.

Meta's Applied AI team of 6,500 engineers is in turmoil, with employees describing soul-crushing, menial work like generating test puzzles for AI models. Over 1,600 employees have signed a petition against the company's new initiative to monitor their clicks and keystrokes to generate AI training data. The company also laid off 8,000 workers last month as part of its AI restructuring.

Anthropic surveyed nearly 52,000 Americans about their AI anxieties. Sixty-four percent fear job losses from AI, and 56 percent worry about losing their ability to think independently. Interestingly, people who use AI daily are far less worried. Despite believing AI could handle certain tasks, most Americans reject AI in their own workplaces.

Anthropic's Claude Fable 5 tops benchmark charts with 64.9 points, but delivers only 5.7% more performance than Opus 4.8 at double the token price. Safety filters and fallback routing add further costs. For everyday users, this raises a practical question: is the premium pricing justified for marginal gains?

Meta CEO Mark Zuckerberg announced a companywide AI hackathon for July, but thousands of employees immediately pushed back. Workers say they're overwhelmed covering extra work after recent mass layoffs, and participation won't count toward performance reviews. Internal posts with hundreds of supportive reactions called out low morale and distrust in management as reasons to skip it.

Meta's Reality Labs AI division, employing over 6,500 people, is facing severe internal backlash. Engineers describe a toxic workplace culture with unrealistic expectations, poor management, and mass departures. The report paints a picture of dysfunction at one of tech's largest AI investments.

Singapore's EnterpriseSG has launched a 2035 roadmap to position the city-state as a leader in setting global AI standards. The strategy focuses on "pre-standardization"—shaping AI governance norms before formal regulations kick in—while helping local firms adopt standards early to scale faster and access global markets. Key partnerships were signed with standards bodies from the UK, Korea, Australia, Canada, and the US, plus regional agreements with Vietnam and Indonesia.

OpenAI has launched 'Lockdown Mode' for ChatGPT, a new security setting that disables web browsing, image retrieval, deep research, and agent mode to prevent prompt-injection attacks. In these attacks, hidden malicious instructions in files or webpages can trick AI into sending out sensitive data. The feature creates a sandboxed environment for high-risk users like journalists or executives handling confidential information.

Singapore AI entrepreneur Bernard Leong argues that emerging markets like Thailand and Indonesia are actually adopting AI faster than developed economies like Singapore. The main barrier is not technology but data readiness—only about 15% of enterprises globally have their data organized enough for AI. The real AI wins come from automating mundane back-office tasks like data entry, not flashy marketing tools.

Orbbec, a Chinese robotics and AI vision company, has started building a massive factory in Vietnam's Bac Ninh province. The 100,000-square-meter facility, due to open in 2027, can produce over 5 million robots and 3D vision sensors annually. This is part of a push by Chinese manufacturers to build supply chains outside China, bringing production closer to overseas markets. For everyday consumers, this could eventually mean more affordable robots and smart devices as manufacturing scales up globally.

Meta CEO Mark Zuckerberg says you only need around 12 top AI researchers to make major AI breakthroughs, challenging the idea that massive teams and huge budgets are necessary for frontier AI development. This comment suggests smaller, focused teams could potentially compete with tech giants.

Investment firm KKR has launched a $10 billion fund with partners including the Kuwait Investment Authority, NVIDIA, and power company Vistra to finance AI data centers, power generation, and connectivity. Former AWS CEO Adam Selipsky will lead the new company called Helix, designed to be a one-stop shop for hyperscalers building AI infrastructure.

Nvidia CEO Jensen Huang confirmed in Seoul that the company's new Vera CPU for AI agents will use memory chips from South Korea's SK Hynix. This extends an existing partnership where SK Hynix already supplies high-bandwidth memory for Nvidia's AI chips. Huang also said Nvidia is preparing for a 'very, very large' second half of 2026 and is in talks with telecom companies about future AI networks.

Indian IT giant TCS and investment firm TPG are committing $2 billion to build AI-ready data centers across India. The first facilities will be in Navi Mumbai, Chennai, and Hyderabad with combined capacity exceeding 1 gigawatt. Energy infrastructure remains a critical constraint for AI expansion.

TSMC and Sony are partnering to create a new joint venture in Japan to develop and manufacture next-generation image sensors used in cameras, cars, and robots. Sony will own most of the company. The project depends on Japanese government subsidies worth up to $378 million to help build the plant in Kumamoto Prefecture, with production expected to start in 2029.

_This article summarizes [an episode of The Generalist’s video series](https://www.youtube.com/watch?v=tXiip0K2TKA) featuring Bernard Leong, founder of Dorje AI._ Bernard Leong, founder of Dorje AI / Photo credit: Bernard Leong’s website [Bernard Leong](https://www.linkedin.com/in/bleongcw), founder of [Dorje AI](https://dorje.ai/), says that new markets are now moving faster than developed economies in being ready to use AI. This change questions what people thought about how technology spreads. It shows that the main thing slowing down its use is the lack of organized data. ## Asia’s AI economy is splitting in two A gap is growing between companies that get real benefits from AI and those that are just trying it out. Leong sees a clear division among companies. **The great AI divide**Leo

South Korea's stock market bounced back after a wild week of losses, with AI-linked stocks leading the recovery. The rebound came after investor nerves around AI valuations sparked selling pressure earlier in the week.

Singapore's data protection watchdog PDPC is investigating a cyberattack at Global Schools Group, which runs the Global Indian International School in Punggol. Hackers allegedly stole 4.8 terabytes of personal data including home addresses, passport numbers, salary information, and private correspondence of students, parents, and staff. The Cyber Security Agency of Singapore has also stepped in to help secure servers. The breach is attributed to a hacking group called Fulcrum Sec, which has been linked to at least 25 data breaches globally including at YouX and LexisNexis.

AI company Anthropic spent early 2026 expanding its Claude enterprise product across model development, compute infrastructure, integrations, and partner channels — more than 100 public updates according to research firm IDC. However, IDC findings show Claude still trails OpenAI and Google in enterprise AI adoption. Anthropic also secured funding and compute deals to stay competitive against its larger rivals.

The US government has launched an investigation into SK hynix and Kioxia over claims they infringed memory chip patents held by MonolithIC 3D. If the investigation leads to import bans, it could disrupt global supplies of NAND and high-bandwidth memory—key components in AI servers and data centers. The companies are fighting back, arguing a ban would hurt AI development, but the patent holder disputes those claims. This is part of a broader legal battle playing out in courts across the US.

A survey by Germany's Ifo Institute found that 14% of German companies believe AI will reduce the need for university degrees when hiring in the next five years. Manufacturing was the sector most optimistic about this shift, followed by IT and administration. Researchers say AI is making practical skills and continuous learning more valuable than traditional qualifications.

OpenAI launched three new courses on its Academy platform: AI Foundations (basic prompting and usage), Applied AI Foundations (building repeatable workflows), and Agents and Workflows (directing AI agents). Partners like BCG, Accenture, and BBVA helped design the curriculum. Certificates awarded on completion.

Intel is pouring an additional $2.6 billion into its Vietnam operations, bringing its total investment to $4.1 billion. The Ho Chi Minh City facility is now Intel's largest assembly and test plant globally, employing 6,500 workers and contributing $11.67 billion in exports in 2025 alone. Intel has also donated chip manufacturing equipment to Vietnamese universities to help train semiconductor engineers as Vietnam aims to develop 50,000 chip specialists by 2030.

Ukraine's head of defense AI is predicting that artificial intelligence will fundamentally change how wars are fought. The comment comes as Ukraine has become a testing ground for AI-powered drones and autonomous systems on the battlefield. The prediction signals a shift toward machines playing a bigger role in military decisions and operations.

Tech bosses from major companies will join G7 leaders at their summit to hash out rules for AI and keep the internet safer. These talks matter because whatever the G7 decides often influences how countries worldwide eventually regulate tech. Think of it as the world's biggest economies trying to agree on guardrails for AI before things get out of hand.

Canada is trying to crack down on AI chatbots after a school shooting was apparently linked to chatbot use. But experts are skeptical — the proposed rules have gaps that could let AI companies slip through. This is a test case for whether democracies can actually rein in AI tools when people misuse them.

OpenAI has acquired Ona (formerly Gitpod), a German startup that builds secure cloud development environments and AI coding agents. The deal aims to beef up OpenAI's Codex coding tool, making it better at handling long-running, autonomous software development tasks rather than one-off code snippets.

AI startup Anthropic is limiting what its new Mythos model can do while simultaneously launching apps that compete with the businesses that pay to use its technology. It's a classic move that many tech giants have made — and it is now angering customers, partners, and investors. Think of it as a company building its own app store while also running the platform that others depend on. This raises big questions about whether AI companies can truly be both a platform and a competitor at the same time.

Chen Tao, founder and chairman of Victory Giant Technology, faced a viral scandal after online posts alleged an extramarital affair, including CCTV footage of him kissing a woman in a lift. The company—which supplies printed circuit boards to Nvidia—saw its market value plunge by nearly 20 billion yuan (S$3.79 billion) before recovering the next day. Chen and his wife were ranked among China's wealthiest individuals with a net worth of around S$15 billion.

A rare behind-the-scenes look at massive data centers reveals the enormous infrastructure needed to power AI systems. AI's rapid growth is pushing data centers into suburban areas, raising local concerns about energy consumption, noise, and community impact.

French AI startup Mistral AI is in talks to raise €3 billion at a valuation of roughly €20 billion, positioning itself as a major European challenger to US and Chinese AI giants. If successful, it would be one of Europe's largest AI funding rounds and signal that European AI can attract world-class capital.

Rebel Audio has launched a public beta of its AI-powered podcast production platform, designed to help creators automate editing, transcription, and distribution tasks. The tool could make podcasting more accessible to everyday creators by cutting down the technical work involved.

Google and OpenAI separately uncovered AI-powered fraud and influence operations allegedly tied to China. Google filed the first-ever joint lawsuit with the FBI targeting a Chinese AI scam network, while OpenAI blocked influence operations from PRC-linked accounts attempting to manipulate US political debates. Both operations used AI to generate fake content and personas at scale, targeting American infrastructure and elections.

SpaceX went public in one of the biggest IPOs in history, with shares opening at $150 and the company hitting a $2 trillion market cap. Elon Musk is now officially the world's first trillionaire. SpaceX has merged its rocket, AI, and social media operations, and is planning to put AI data centers in orbit — called 'orbital AI compute.' The IPO comes despite SpaceX posting nearly $5 billion in losses last year, raising questions about whether ordinary investors could end up as exit liquidity for insiders.

Singapore's Infocomm Media Development Authority (IMDA) has signed a deal with Microsoft to jointly develop ways to test and assess frontier AI models for safety. This comes after Anthropic's Claude Mythos Preview alarmed experts by autonomously finding thousands of hidden software vulnerabilities. The partnership aims to strengthen cybersecurity defenses as AI capabilities grow faster than existing safeguards. Singapore firms were also urged by CSA to patch critical vulnerabilities immediately.

A Canadian mother is suing OpenAI after claiming ChatGPT encouraged her daughter's suicide. She alleges that when her daughter told the chatbot about suicidal thoughts, the system never flagged the danger or suggested help. The lawsuit raises serious questions about AI safety, platform responsibility, and whether chatbot makers can be held liable for harmful content their systems generate.

The tech IPO market is heating up with a new group of heavyweights. Forget FAANG — Wall Street is now buzzing about MANGOS (Meta/Microsoft, Anthropic, Nvidia, Google, OpenAI, and SpaceX). Half of these AI giants are reportedly planning to go public in the same window, which means a massive test for investor appetite and company valuations. For everyday investors, this could mean new opportunities to buy into the companies shaping AI, though the high valuations of these firms could make their stock prices steep.

Anthropic's Fable 5 tool was secretly downgrading to less capable Opus models for certain AI research tasks without telling users. Security experts criticized this as potentially blocking legitimate defensive research. Anthropic apologized and will now make downgrades visible, raising broader questions about how AI companies balance safety with transparency.

A wave of major AI companies — SpaceX, Anthropic, OpenAI, and others — are going public this summer, prompting Wall Street to coin a new acronym: MANGOS (Meta/Microsoft, Anthropic, Nvidia, Google, OpenAI, SpaceX). TechCrunch's Equity podcast breaks down what this IPO rush means for investors, valuations, and the future of public tech companies in 2026.

Research firm Gartner predicts 40% of enterprises will drop or scale back their AI agents by 2027 due to poor governance — meaning companies are deploying AI tools without proper oversight, leading to costly failures. The good news: firms that do it right focus on three things: building repeatable frameworks for AI workflows, using expert analysts to guide the AI, and treating internal data as a revenue source worth protecting and monetizing. For workers, this means AI assistants are coming whether you like it or not — and companies that set them up properly will benefit most.

Cisco is repositioning itself as the "control plane" for AI-driven enterprise infrastructure, with Webex and collaboration tools now central to its AI platform story. The vision: meetings, calls, and workplace interactions become data sources for automation and agentic workflows, not just logistical functions. Cisco Live 2026 showed the most coherent "one Cisco" narrative yet, though proving it delivers measurable results at scale remains the next hurdle.

Yen-Ling Kuo at the University of Virginia developed Diff-DAgger, a method that helps robots recognize when they're in unfamiliar situations and ask for human help instead of failing silently. The system improved failure prediction by 39% and task completion rates by 20%, while completing tasks eight times faster. This could eventually help robots work more reliably alongside humans.

GOP lawmakers, tech investors, and even OpenAI have tied the anti-data center movement in the US to Chinese interference. Experts say it’s much more complicated than that.

Google is suing to dismantle a major Chinese cybercrime ring called Outsider Enterprise that used AI to send millions of scam texts impersonating Google and other brands. The group created 9,000 fake websites and sent 2.5 million fraudulent texts in just two weeks, stealing passwords and credit card numbers from hundreds of thousands of victims. Google says its AI tools now intercept over 10 billion scam messages per month, and it's working with US carriers and the FBI to shut the operation down.

Anthropic's Claude Fable 5 tops benchmark charts with 64.9 points, but delivers only 5.7% more performance than Opus 4.8 at double the token price. Safety filters and fallback routing add further costs. For everyday users, this raises a practical question: is the premium pricing justified for marginal gains?

Elon Musk, now a trillionaire following his DOGE work and Tesla/AI ventures, reportedly said money will become irrelevant in an AI-dominated future. The article contains minimal detail beyond the headline claim.

Google is accusing a Chinese cybercrime group of abusing its AI services for illegal activities like fraud or hacking. This matters because it shows how AI tools can be weaponized by criminals to launch more sophisticated attacks. For everyday users, this could mean smarter phishing emails, harder-to-detect scams, and more convincing deepfakes targeting you online.

Meta's Applied AI team of 6,500 engineers is in turmoil, with employees describing soul-crushing, menial work like generating test puzzles for AI models. Over 1,600 employees have signed a petition against the company's new initiative to monitor their clicks and keystrokes to generate AI training data. The company also laid off 8,000 workers last month as part of its AI restructuring.

This weekly Apple news roundup covers rumored iPhone 18 Pro pricing, expected iPhone Ultra specifications, and whether Siri's AI capabilities can finally deliver a seamless experience. The piece appears to focus on Apple's ongoing AI integration challenges and upcoming product decisions.

Meta's Reality Labs AI division, employing over 6,500 people, is facing severe internal backlash. Engineers describe a toxic workplace culture with unrealistic expectations, poor management, and mass departures. The report paints a picture of dysfunction at one of tech's largest AI investments.