Singapore AI News & Daily Briefing

Singapore has launched the ASPIRE 2B supercomputer, a major upgrade to its national computing infrastructure with over 1,500 Nvidia H200 GPUs delivering up to 115 petaFLOPs of performance — roughly 100 times more powerful than its predecessor. The system will help local researchers train larger AI models, run climate simulations and develop quantum computing applications without relying on overseas infrastructure. Minister Josephine Teo highlighted that over 70 companies have established AI Centers of Excellence in Singapore, contributing to both research and industry deployment.

Singapore has launched a new programme to prepare university students for AI roles in banking and finance, a sector that contributes about 14% of the country's GDP and employs around 200,000 people. The Young Talent Programme for AI in Finance combines coursework with real industry projects and internships at major banks including JPMorgan, DBS, OCBC and UBS. Students will learn to use tools like Bloomberg terminals and deploy AI solutions on cloud platforms, focusing on areas like regulatory compliance, investment research and risk analysis. The pilot kicks off in August 2026 with around 100 students.

PatSnap, a Singapore-founded patent intelligence platform, has grown into a global unicorn by helping companies track patents and avoid infringement. The company now operates across 50 countries and crossed US$100 million in annual recurring revenue in 2024 while turning profitable. Founder Jeffrey Tiong has embraced AI throughout the company — employees get around US$1,000 monthly in AI tokens and compete in monthly innovation challenges. PatSnap uses its clean, curated patent database to train AI models that help R&D teams spot cross-industry innovations, moving beyond traditional patent searches to broader research intelligence.

Tin, a critical material in solder used to assemble AI server components, is facing a supply crunch as global data center construction accelerates. Demand from AI servers is projected to triple by 2030, yet global supply remains concentrated in a handful of politically unstable mining regions. Tin prices have surged over 55% year-on-year, making this one of the least-discussed but most consequential commodity stories of the AI infrastructure buildout.

Google DeepMind just announced a sweeping national partnership with Singapore to deploy frontier AI across healthcare, education, and workforce training. The deal includes AI tools for doctors to support patient care, pandemic preparedness research using AlphaFold, a running assistant for blind athletes built with SG Enable, and free Gemini for Education for all Singapore teachers. Google projects this could generate S$3.3 billion in economic value for Singapore by 2040 through faster R&D. The partnership also involves IMDA working on safety benchmarks for local languages.

Google DeepMind is launching a startup accelerator program in Asia Pacific focused on using AI to tackle environmental challenges like climate change, pollution, and conservation. The program would mentor and fund early-stage AI companies working on these problems.

Google has redesigned its iconic search box for the first time in 25 years, transforming it from a simple keyword input into an AI-powered conversational interface. The new box accepts text, images, PDFs, videos, and even open Chrome tabs. Google is also merging its AI Overviews and AI Mode features into one seamless experience, so users no longer have to choose between traditional results and AI-powered answers. With over a billion monthly users on AI Mode alone, this is the most significant change to how billions of people access information in a generation.

South Africa sits on roughly 88% of global platinum-group metal reserves critical to semiconductor manufacturing, yet its draft AI policy was withdrawn after citing AI-hallucinated references. The country is caught between Chinese and US tech giants—Huawei pitching DeepSeek bundles at 90% lower cost, Microsoft committing $300 million to cloud infrastructure—competing to control Africa's largest data center market. But South Africa has proposed no minimum terms for foreign investment, no data sovereignty requirements, no technology transfer conditions. A new panel will revise the policy by January 2027. Experts warn that infrastructure decisions made now determine what remains renegotiable later, and the answer is: very little.

Hackers have hit Microsoft repositories for the second time in weeks, planting malware in 73 packages that steal cloud credentials the moment developers open them in AI coding agents like Claude Code, Cursor, and VS Code. The malware, called Miasma, harvests login tokens for AWS, Azure, GCP, and Kubernetes, then spreads to other machines. Microsoft only acknowledged the breach days after GitHub blocked the packages. Developers who used AI agents on these packages should assume their systems are compromised.

OpenAI has filed a confidential S-1 registration with the US Securities and Exchange Commission, signaling it is preparing for a potential public stock offering. The company says timing is not yet set and it may stay private longer since some goals are easier to achieve as a private company, but the filing gives it the option to go public sooner if needed.

IBM Research argues that raw AI language models alone aren't enough for businesses — you need 'agent logic,' essentially software that guides AI toward the right answers instead of letting it hallucinate freely. In tests, adding knowledge graphs and program analysis tools to AI cut costs by up to 15 times while improving accuracy for tasks like analyzing old mainframe code, generating software tests, and fixing bugs. IBM says this hybrid approach lets businesses deploy AI that actually works in complex, regulated environments.

OpenAI launched a new platform called the Economic Research Exchange to fund outside researchers studying how AI affects jobs, businesses, and the economy. Researchers can apply now with project proposals, and selected teams will get access to OpenAI tools and data to study questions like AI's impact on wages, productivity, and inequality. Applications close July 5, 2026.

A Cambridge professor is using Google DeepMind's Co-Scientist AI tool to hunt for molecular switches that cause severe diseases like sepsis when viruses jump from animals to humans. The AI sifted through published research and generated hypotheses the scientist hadn't considered, helping narrow down which proteins and specific amino acids to test. This cut what would normally take two to three years of lab work down to roughly six months.

Railway, a cloud platform used by 2 million developers, raised $100 million to build AI-native infrastructure that deploys apps in under one second — compared to the 2-3 minutes it takes with traditional tools like Terraform. The company abandoned Google Cloud and built its own data centers to keep pace with AI coding assistants that generate code in seconds. Railway claims 10x faster developer velocity and 65% cost savings compared to AWS, with some enterprise customers reporting 87% cost reductions. 31% of Fortune 500 companies already use the platform.

Salesforce has rebuilt Slackbot from a basic notification tool into a full AI agent that can search enterprise data across Salesforce, Google Drive, and Slack conversations, draft documents, and take action on behalf of employees. Powered by Anthropic's Claude, it's being positioned as a central "super agent" that coordinates AI across the workplace. In internal testing with 80,000 employees, two-thirds tried it and 80% kept using it, with satisfaction rates at 96%. Employees report saving 2-20 hours per week. Slackbot is included free for Business+ and Enterprise+ customers.

Anthropic just launched Cowork, a desktop AI agent that non-technical users can command to organize files, create expense reports from receipt screenshots, or draft documents from scattered notes — all without touching code. The AI gets access to a specific folder on your computer and can read, edit, or create files there. Built in just 10 days (reportedly using its own AI coding tool), it's currently exclusive to Claude Max subscribers on macOS at $100-200 per month. Anthropic is openly warning users the AI could delete files if it misinterprets instructions.

Anthropic's Claude Code charges up to $200 per month for AI coding assistance, but there's now a free alternative. Block (formerly Square) built Goose, an open-source AI agent that runs entirely on your local machine — no subscription, no cloud dependency, no rate limits. Developers can connect it to open-source models like Qwen and run it offline, even on an airplane. The project has over 26,000 GitHub stars and 362 contributors. The trade-off: local models aren't quite as capable as Claude Opus for the most complex coding tasks.

OpenAI published a wide-ranging manifesto laying out its vision for AI's future. The company argues AI should be widely distributed to benefit everyone, not concentrated among a few powerful players. Key goals include building an AI that can conduct research autonomously, accelerating economic growth so gains are shared broadly, and eventually giving every person access to personal AGI. OpenAI believes most of its research could be done by AI systems by March 2028.

AI startup Majestic Labs is building a server called Prometheus with up to 128 terabytes of memory—over 60 times more than Nvidia's top AI server. The company argues current AI systems are 'memory-starved' because moving large AI models in and out of memory is slow and expensive. Prometheus uses standard DRAM with custom chips to connect memory across longer distances, paired with custom AI processors. The company claims this could cut customers' costs and power usage by 10 to 50 times. Servers are expected in 2027.

Google has released Gemini Omni Flash, a new AI model that creates and edits videos simply by chatting with it. Users can input images, text, audio, or existing video clips, then ask the AI to transform scenes — like turning a sculpture into bubbles or adding ripple effects. It's now available on YouTube Shorts, Google Flow, and the Gemini app for subscribers.

Google DeepMind's Co-Scientist AI helped biologists at MIT accelerate aging research by months. The AI scanned tens of thousands of scientific papers and proposed novel genetic factors that push aging cells back toward a youthful state. What normally takes researchers six months to analyze now takes a few days. The AI validated its own hypotheses in lab tests, successfully identifying factors that improved cell function.

One-third of the world's internet users are under 18, yet digital systems were built for adults. IEEE's president argues that protecting children online is an engineering challenge, not just a policy debate. Governments in the EU, UK, Australia, Indonesia, and the US are now mandating age-appropriate design—but technical standards from bodies like IEEE are helping bridge the gap between policy goals and actual engineering practice.

Google has released Antigravity 2.0, a standalone desktop app for working with AI agents. Unlike the previous IDE version focused on coding, this new release expands to knowledge work — agents can schedule tasks automatically, call subagents for different jobs, and integrate with JSON systems. It's designed for businesses and developers wanting to automate complex workflows.

JetBrains released Mellum2, a 12-billion parameter open-source AI model built specifically for production systems. Unlike traditional models that use all parameters for every query, Mellum2 activates only 2.5 billion at a time — making it over twice as fast as comparable models while delivering similar results. It's designed as a 'workhorse' model for tasks like routing requests between AI systems, summarizing search results, or handling smaller jobs in an AI pipeline without the cost of calling a massive frontier model.

Google has launched a suite of AI tools specifically for scientists, called Gemini for Science. These tools aim to help researchers analyze data, run experiments, and accelerate discoveries across fields like biology, chemistry, and physics. The initiative combines Google's AI capabilities with scientific workflows to expand what researchers can accomplish.

Google is rolling out tools to help people verify whether videos and images were created or edited using AI. The system uses invisible watermarks called SynthID that get embedded in AI-generated content, allowing users to check authenticity through Google Search, Chrome, and the Gemini app. This addresses growing concerns about deepfakes and misleading AI content.

A new AI system combining thermal cameras and machine learning is protecting gray whales in San Francisco Bay from deadly ship strikes. Climate change has diverted whales off their traditional migration route, and ship strikes now cause 40% of bay deaths. The WhaleSpotter system detects whale spouts (warm exhalations visible in thermal footage) from a Coast Guard tower, verifies detections with humans, then alerts nearby ships to slow down. Already logging 6,600 detections in its first week, the system claims a 90% reduction in strike risk. Human verifiers currently confirm detections to prevent alert fatigue, but the company aims to automate fully.

Google DeepMind has launched Project Genie, a tool that uses nearly 20 years of Street View photos to create realistic virtual worlds. The AI studies real locations and generates new environments based on them. This could change how video games, virtual reality experiences, and simulations are built — creators could 'visit' real places in digital form without physically going there.

Scientists have developed a radar system that uses machine learning to identify different pollinator insects—like bees and wasps—without having to catch or harm them. By analyzing the tiny wobbles in radar signals caused by insect wingbeats, the system achieved 85% accuracy in telling species apart. This could replace traditional insect traps that kill pollinators with cyanide, and eventually help monitor bee populations and track invasive species globally.

MathWorks is hosting a free webinar on designing AI-based virtual sensor models using MATLAB and Simulink. The workflow covers training, validation, compression for embedded deployment, and generating C code for hardware testing. It's targeted at engineers working on sensor integration in system-level simulations.

An experienced chip designer who spent decades in academia shares what he learned moving to industry: the goals are completely different. Academia asks 'what's possible,' while industry asks 'what works reliably at scale.' At advanced chip nodes, a single mask set can cost tens of millions of dollars, so first-time success is mandatory. The piece explains why companies rely heavily on pre-built silicon intellectual property from firms like Arm, Cadence, Synopsys, and Rambus rather than designing everything in-house. The ASIC market is projected to grow from $23.4 billion to $38.8 billion by 2033, driven by demand from automotive and AI applications.

The term 'social engineering' now triggers thoughts of hacking and phishing, but its roots lie in deliberately shaping human behavior at scale. This thought-provoking essay traces how the concept evolved from 19th-century workplace reforms to today's algorithm-driven manipulation through social media. Modern tech companies use behavioral nudges, personalization, and engagement algorithms that subtly shape our choices without explicit consent.

TechCrunch has followed SpaceX's start, struggles, and successes from the early days. And we're here for what happens next too. This package of SpaceX IPO coverage includes who stands to win (and maybe some who won't), pre-IPO deals, and what's tucked inside its S-1 registration document.

Former US presidential candidate Andrew Yang is pushing a new startup idea: businesses that lower the cost of living instead of extracting money from customers. His latest venture, Noble Mobile, offers budget cell service and gives customers money back if they use less data. Yang argues that as AI threatens to displace workers, startups focused on affordability in housing, food, healthcare, and wireless could become a major opportunity.

A coalition of US state attorneys general, led by New York, has opened an investigation into OpenAI and issued a subpoena covering everything from advertising practices and user data handling to how it treats minors and seniors. This comes on top of an existing Florida lawsuit alleging OpenAI ignored safety warnings and endangered children. With OpenAI having just filed for IPO, the timing puts intense scrutiny on the company's practices right as it seeks public investment.

Over half of Americans—53%—worry AI could take their job or a family member's job, according to a new Reuters/Ipsos poll. The fear is backed by real layoffs: companies like Intuit cut 17% of staff while citing AI. AI was behind a quarter of the 60,620 US layoffs announced in March. Even the Pope has warned about AI's societal risks, and University of Arizona students recently jeered former Google CEO Eric Schmidt over AI at graduation.

Spain could lose a net 400,000 jobs to AI between 2025 and 2035, according to research from the Funcas Finance and Technology Observatory. While AI creates 1.61 million new jobs and benefits 3.1 million workers through complementarity effects, it displaces 2 million roles—mostly in administrative, technical, and scientific fields. Mid-level white-collar workers face the biggest risk, reversing previous automation trends that hit manual labor hardest.

Meta is unwinding its $2 billion acquisition of Manus, an AI agent startup, after Beijing demanded the deal be reversed. Manus had developed AI agents capable of automating complex computer tasks. The reversal highlights how Chinese government approval remains critical for tech acquisitions involving Chinese companies.

# Škoda’s New EV Will Likely Be Its Most Expensive Yet | WIRED Social Media - [x] On They are capable of tracking your browser across other sites and building up a profile of your interests. This may impact the content and messages you see on other websites you visit. If you do not allow these cookies you may not be able to use or see these sharing tools. * * * Essential - [x] On This website uses essential cookies and services to enable core website features and provide a seamless user experience. These cookies and services are used to facilitate features such as navigation, remember user preferences, and ensure the security of the website. * * * Targeted - [x] On These cookies may be set through our site by our advertising partners. They may be used by those companies to build a profile

# I found the most effortless way to clean your Android phone storage - and it's free to use | ZDNET X Trending * [I wore the world's first HDR10 smart glasses](https://www.zdnet.com/article/rayneo-air-4-pro-hdr10-smart-glasses-ces/) * [TCL's new E Ink tablet beats the Remarkable and Kindle](https://www.zdnet.com/article/tcl-note-a1-nxtpaper-ces/) * [Anker's new charger is one of the most unique I've ever seen](https://www.zdnet.com/article/anker-wireless-chargers-ces-2026/) * [Best laptop cooling pads](https://www.zdnet.com/home-and-office/best-laptop-cooling-pad/) * [Best flip phones](https://www.zdnet.com/article/best-flip-phone/) * [Roku TV vs Fire Stick](https://www.zdnet.com/home-and-office/home-entertainment/roku-tv-vs-fire-stick/) * [Galaxy Buds 3 Pro vs Apple AirPods Pro 3](https:

 Cesar Cadenas/ZDNET _Follow ZDNET:_[Add us as a preferred source](https://cc.zdnet.com/v1/otc/00hQi47eqnEWQ6T9d4QLBUc?element=BODY&element_label=Add+us+as+a+preferred+source&module=LINK&object_type=text-link&object_uuid=61bd6244-1d37-4fef-a429-0e9ee6f45b2b&position=1&template=article&track_code=__COM_CLICK_ID__&url=https%3A%2F%2Fwww.google.com%2Fpreferences%2Fsource%3Fq%3Dzdnet.com&view_instance_uuid=61bbe448-c11d-4ce0-8066-3426e74f5890&object_version=01d50d3d-0771-488b-a080-9aa6e7fc5fc5)_on Google._ * * * ### ZDNET's key takeaways * Motorola's Razr+ challenges the idea that yo

[ ## Ugreen FineTrack 2 4 / 5 Very good](https://cc.zdnet.com/v1/otc/00hQi47eqnEWQ6T9d4QLBUc?element=BODY&module=BUTTON&object_type=review-card%7Cbutton%7C1&object_uuid=37bf64b3-c4e6-4b07-bc6e-9689057c81cd&object_version=757814a4-3433-4123-ad85-01ce77e50cfb&position=1&product_uuid=ed0ebc75-28d9-4f13-b0cd-87f27b07aef4&template=article&track_code=__COM_CLICK_ID__&view_instance_uuid=47a21d44-15ba-44ad-9ddf-32e37e11147d&url=https%3A%2F%2Fassoc-redirect.amazon.com%2Fg%2Fr%2Fhttps%3A%2F%2Fwww.amazon.com%2Fdp%2FB0G64RT15Y%3FlinkCode%3Dogi%26th%3D1%26psc%3D1%26tag%3Dzd-buy-button-20%26ascsubtag%3D__COM_CLICK_ID__%257C47a21d44-15ba-44ad-9ddf-32e37e11147d%257Cdtp) #### pros and cons **Pr

OpenAI just announced a $150 million bet on helping businesses actually use AI. They're launching a Partner Network that connects enterprises with consulting firms who can help them integrate AI into their workflows—not just access the models, but actually get employees to use them. The goal is to train 300,000 certified AI consultants by end of 2026. This matters because most businesses struggle with the messy middle ground of AI adoption: figuring out which tasks to automate, fitting AI into existing systems, and convincing workers to change how they do their jobs. OpenAI is essentially admitting they can't do this alone and needs an army of partners to make AI stick in the enterprise world.

Apple's upcoming iPhone 18 Pro with iOS 27 is being positioned as the 'first true AI device' following its WWDC 2026 announcements. The company showcased new Siri AI features with live demos to prove they work. While some reviewers are excited, others call it a 'me-too product' that just delivers on promises made two years ago. The software won't be available until fall, so real-world performance remains untested.

Companies across Asia Pacific and Japan are moving AI workloads from public cloud services back to private cloud infrastructure faster than anywhere else globally. A Broadcom report found 82% of APJ organizations have considered this shift as AI moves from experimental pilots into real production use. This trend suggests businesses want more control over their AI computing resources and data.

Anthropic is expanding its Glasswing security program to 150 more partners in over 15 countries, including the European Union. The initiative gives vetted organizations access to its powerful Mythos AI to find software vulnerabilities before attackers can exploit them. Since April, partners have uncovered over 10,000 critical bugs. This move comes after the White House raised concerns about the technology being misused.

South Korean tech firms Samsung Electro-Mechanics and LG Innotek posted stronger Q1 earnings, driven by surging demand for AI-related chip components like advanced FC-BGA substrates and multilayer ceramic capacitors. Samsung Electro-Mechanics' substrate demand now exceeds supply by over 50%, prompting plans to more than double capital spending to over US$1.3 billion in 2026, including a new plant in Vietnam. AI servers require 10 to 15 times more of these components than standard servers, creating a seller's market that could eventually raise costs for companies building AI infrastructure.

A police officer in Derbyshire, UK is under criminal investigation for allegedly using AI-generated material as evidence. The case raises serious questions about the integrity of AI-created content in criminal proceedings — if police can be tempted or tricked into using fake AI evidence, the implications for justice and public trust are enormous.

Ant Group, the company behind Alipay, launched LingGuang — a multimodal AI assistant that can build and run custom mini-apps in just 30 seconds using simple chat commands. Users can generate apps for event planning, financial calculators, or other tasks without writing code. The tool handles text, images, voice, and data, outputting results as 3D models, audio, charts, or animations.

Tech in Asia published an opinion piece exploring the broader implications of AI's rapid expansion across industries. The article appears to examine what opportunities or roles remain for humans as AI capabilities grow.

Singapore startup Pints AI has raised $5.6 million to expand its Autothought platform, which helps banks and insurers deploy AI that regulators can actually audit. The platform connects to core financial systems and automatically creates a paper trail for every AI decision—whether it's approving a loan, processing a claim, or onboarding a customer. Twelve institutions across four countries have already saved a combined $10 million using the system. The funding came from Singapore VCs including Tin Men Capital, NTUitive, and SUTD Venture Fund.

Singapore-based venture capital firm Tin Men Capital has invested $5.6 million in Pints AI, an AI-powered startup. This marks Tin Men Capital's continued bet on AI companies in Southeast Asia, where the technology sector is seeing growing investor interest.

ByteDance—the company behind TikTok—is in talks to buy AI chips from China's Iluvatar CoreX, according to sources. This move appears aimed at securing computing power for AI development after US export restrictions limited ByteDance's access to American chips like Nvidia's GPUs. Iluvatar CoreX is a Chinese AI chip startup competing in the space dominated by Nvidia and AMD.

Microsoft CEO Satya Nadella has warned that a small number of dominant AI companies could become so powerful they might destroy entire industries. He flagged concerns about market concentration in the AI sector, suggesting that without competition safeguards, a few winners could use AI to sideline rivals and consolidate control over entire markets.

ByteDance's AI chatbot Doubao is now processing over 120 trillion tokens daily—a massive jump driven by AI video generation and agent tools. The company holds nearly half of China's public cloud AI market, outpacing Baidu and Alibaba combined. New pricing cuts costs by up to 70%, making enterprise AI adoption cheaper.

Tencent Cloud launched an upgraded AIoT platform with voice and video intelligence features for smart devices. New tools like TWeTalk (speech-to-text, emotion recognition) and TWeSee (video analytics) aim to simplify integration for device makers. Partners include makers of Talking Tom AI Bud and Alpha Group, targeting smart toys, robotics, and surveillance.

Schneider Electric and Foxconn have announced a partnership to develop AI data center infrastructure. Schneider Electric brings its power management expertise while Foxconn contributes its manufacturing capabilities. This collaboration aims to address the growing energy demands of AI operations, which require massive computing power and cooling systems. For everyday users, this could eventually mean more reliable AI services, though the direct consumer impact will take time to materialize.

Update on DeepMind partner's dishwashing humanoid fools viewers before founder admits it's fake. Coverage via Yahoo Singapore Tech.

A PwC report suggests AI professionals in Singapore command salaries at least 32% higher than non-AI roles, highlighting the premium placed on AI skills in the local job market. The content snippet focuses on India's AI manufacturing potential, but the headline signals strong demand and pay premiums for AI talent in Singapore—likely driven by financial services, tech firms, and government digital transformation initiatives. For job seekers, this reinforces AI expertise as a high-value career path; for employers, it signals intensified competition for talent.

Samsung spent a record US$25.7 billion on R&D in 2025, up 8% from the prior year, pouring money into next-generation AI chips and memory. The company also invested US$35.9 billion in facilities, building new semiconductor plants including an R&D complex. Samsung is fighting to reclaim ground in high-bandwidth memory (HBM), where rival SK Hynix now leads. Memory now makes up 650% of a data center server's component cost, up from 458% a year ago—pushing up prices for everything from smartphones to cloud services. The spending signals intensifying competition in AI hardware as demand for memory chips explodes.

Databricks just locked in over US$4 billion in fresh funding, pushing its valuation to US$134 billion — one of the biggest funding rounds ever for a data and AI software company. Investors include heavy hitters like GIC and Temasek (Singapore's sovereign wealth funds), Insight Partners, and BlackRock. The company is now generating US$4.8 billion in annual revenue and its AI products alone hit US$1 billion run-rate. The money will fund building AI 'agents' — autonomous software that handles complex business tasks — and potential acquisitions.

Austrian chipmaker AT&S signed a deal with AMD to supply high-end computer chips for AI and supercomputers. The company will invest €1.5-2 billion (US$1.74-2.32 billion) to expand factories in Malaysia and China. These chip substrates — the foundational hardware that AI processors sit on — are critical components in the AI computing race. Malaysia and China facilities will handle the production surge needed to keep up with AI chip demand.

NanoClaw, a security-focused startup that runs AI agents in isolated containers, has raised $12 million in seed funding from investors including Docker, Hugging Face CEO Clem Delangue, and notably, Singapore Foreign Minister Vivian Balakrishnan. The company emerged after its founder discovered that an existing AI agent framework had stored his WhatsApp messages as unencrypted plain text on his computer. NanoClaw now offers a safer alternative using Docker Sandboxes layered inside microVMs to prevent compromised agents from reaching sensitive data on the host machine. The startup has gained massive developer attention with around 22,000 GitHub stars within weeks.

Google has filed a lawsuit against Outsider Enterprise, a Chinese cybercrime group, for using AI to power massive text-message phishing scams that stole an estimated $1.9 billion from victims. The AI was used to automate and scale the scams targeting Android phone users in the US. Google wants the court to shut down the operation and prevent further attacks. This is a reminder that scammers increasingly use AI to sound more convincing — always verify unexpected texts claiming to be from your bank or delivery company.

Databricks released Omnigent, a free open-source tool that helps companies manage and govern multiple AI agents running at the same time. Think of it as a control panel for the growing patchwork of AI systems enterprises now run internally. Released under an open license, any developer can use or modify it. This matters because as AI agents multiply inside companies, keeping them secure, compliant, and working together is becoming a real headache.

Tech in Asia's 60/40 podcast checked in on their January 2026 predictions six months in. Hits: Google AI mode became default, Anthropic emerged as a serious competitor, Canva is headed toward a $60 billion IPO. Misses or concerns: questions linger about whether AI startup revenue is inflated by trials rather than real paying customers, and developers using AI coding tools may be shipping code they don't fully understand. The reckoning for inflated ARR likely comes after the first major AI IPO.

Your AI Model Can Disappear Overnight — a sobering look at the fragility of relying on third-party AI services. Without direct access to model weights, data, or infrastructure, businesses can find themselves stranded if a provider shuts down, changes pricing, or deprecates a model. The piece likely explores contingency strategies for enterprises.

A US order forced Anthropic to shut down its Fable 5 and Mythos 5 AI models worldwide. European researchers are now debating how to respond — whether to build homegrown AI foundation models or negotiate contracts for access. Experts warn Europe currently lacks the computing power, energy infrastructure, and competitive providers needed to build independent AI systems, raising questions about tech sovereignty.

Cybersecurity experts are pressuring the US government to remove restrictions on Anthropic's security-focused AI models. They're arguing that blocking these models could actually weaken national defenses by limiting access to advanced AI tools that help fight cyber threats. This debate about balancing AI safety rules with real-world security needs is heating up.

Anthropic pulled its Fable 5 and Mythos 5 AI models for everyone worldwide after the US government cited national security concerns about potential jailbreaking. The company had just 90 minutes to disable access for all users, including its own employees. Anthropic disputes the narrow technical finding and is working to restore access.

Keeper Security launched a new feature called Universal Secrets Sync that automatically rotates and syncs passwords across AWS, Azure, and Google Cloud simultaneously. This closes the dangerous gap between stored credentials and what's actually running in production systems — a common source of breaches.

Tensordyne, a chip startup, is expecting $200 million in orders for its AI computing system designed to compete with Nvidia. The company is positioning its hardware as an alternative for customers looking to reduce their dependence on Nvidia's dominant GPU market. While impressive for a startup, this remains a niche development in the massive AI chip industry dominated by established players.

New data shows US shoppers who click through AI-powered recommendations browse longer and spend more per visit compared to regular shoppers. This validates how AI recommendation engines are becoming a key revenue driver for retailers.

Microsoft shareholders are suing the company over spending concerns, targeting its cloud business and AI investments. The lawsuit suggests investors believe too much money is being poured into AI infrastructure without clear returns.

Nvidia is selling at least $20 billion in bonds—its first bond sale since 2021. This is part of a broader trend where AI companies are taking on debt to fund massive infrastructure spending. The chip giant joins other AI firms tapping credit markets to finance data centers and hardware buildouts. Bloomberg reports this signals just how capital-intensive the AI race has become.

Update on US saw risk of Anthropic models being diverted to foreign military intelligence. Coverage via Yahoo Singapore Tech.

The US government ordered AI company Anthropic to shut down access to its most powerful models for all users worldwide, including its own employees. The move sparked immediate backlash and is being seen as a wake-up call for countries relying on American AI companies. Now there's fresh pressure on nations to build their own AI systems rather than depend on US firms that can be pulled offline at Washington's request.

The US government ordered Anthropic to shut down access to its newest and most powerful AI models, Fable 5 and Mythos 5, for all foreign nationals—including the company's own employees. The order came after Amazon flagged security concerns about potential misuse. Anthropic complied but disagreed with the decision, saying the vulnerabilities found were minor and available in other models. The unprecedented move left hundreds of millions of users without access to these tools overnight.

Over 100 cybersecurity executives are pushing the US government to reverse restrictions on Anthropic's most advanced AI models. The company took its Mythos and Fable 5 models offline to comply with a directive that blocks foreign nationals from accessing the technology. Experts argue the move could backfire — China's AI is 'only months behind' American models, and restricting top cyber defense tools may actually help adversaries more than hurt them. The dispute stems from Anthropic's refusal to let the Pentagon use its technology for autonomous weapons and domestic surveillance.

Manulife Hong Kong has joined a regulator-backed program that brings insurance companies and tech firms together to share best practices on using AI responsibly. The Insurance Authority's AI Cohort Programme aims to help the insurance sector adopt AI tools safely and effectively. For everyday policyholders, this could mean faster claims processing and more personalized coverage options as AI becomes more integrated into insurance operations.

A humanoid robot will soon run a convenience store in Hong Kong, marking another step in AI-powered automation entering everyday retail spaces. While robotic helpers have appeared in warehouses and hotels, a fully robot-staffed shop is still rare. For shoppers, this could mean 24/7 service without human staff — though whether customers actually prefer interacting with robots over people remains to be seen.

Singapore startups are rapidly expanding their use of AI tools, with 42% more companies subscribing to AI platforms in 2026. Most now run multiple AI services simultaneously, with ChatGPT leading but Anthropic's Claude closing in fast. Claude users spend 1.4 times more per account, suggesting they rely on it for heavier workloads.

Singapore's UOB bank and NUS-ISS are shaping how workers adapt to AI. UOB's HR head says most jobs will eventually be affected by AI, but workforce cuts at the bank had "nothing to do with AI" - they were about process redesign. The bank runs a "Better U Pivot Programme" with an 80% success rate helping at-risk employees transition to new roles. NUS-ISS's deputy CEO argues the real shortage isn't just AI specialists, but "AI bilinguals" - people who keep their domain expertise and learn to apply AI within it. Both leaders say problem-solving, curiosity, and adaptability matter more than technical coding skills when hiring.