Singapore AI News & Daily Briefing

NUS is now using AI-powered digital assessment tools to train clinical students. The system can evaluate student performance, provide feedback, and track competency development in medical training — potentially complementing traditional clinical rotations. This comes as Singapore pushes to modernize healthcare education through technology.

AI tasks like image recognition and real-time translations are now running on mobile phones, but heavy AI workloads are exposing gaps in APAC mobile networks—particularly upload speeds and delays. Singapore has the best latency in the region, while Malaysia shows steady performance. However, cloud routing issues and network jitter remain problems across most markets. This matters because slower AI on phones means laggy translations, poorer video calls, and apps that don't work as well on mobile data.

NTU Singapore is partnering to develop AI systems that can manage satellite fleets more efficiently. The technology aims to help coordinate multiple satellites, optimize their operations, and reduce the need for human ground control. This could make space operations cheaper and more reliable, though practical applications for everyday users are still years away.

Singapore food-tech startup WhyQ, once burning cash on hawker delivery, has pivoted to B2B corporate dining and hit profitability. Now it's launching 'WhyQ Intelligence'—an AI tool that tracks employee nutrition, suggests meals, and connects food choices to workplace wellness data. The company aims for 70-80% employee engagement at client firms. It plans to expand to Hong Kong and Sydney once it secures anchor corporate contracts.

Enterprise AI spending is spiraling out of control, warns ManageEngine's CEO. The company is betting that businesses will prefer efficient, targeted AI solutions over massive, expensive frontier models. ManageEngine argues companies can get more value from AI by solving specific problems cheaply rather than chasing raw power. This reflects a growing industry split between those building massive AI infrastructure and those trying to cut costs.

Google has launched Gemini 3.5 Live Translate, a real-time voice translation tool that works across over 70 languages. Unlike traditional translation apps that wait for you to finish speaking, this one translates continuously with natural-sounding speech that preserves your tone and pace. It can be used in Google Translate on your phone, in Google Meet for business calls, and — notably — Grab is testing it to help drivers and riders communicate across language barriers during the 10 million voice calls they handle monthly.

Hugging Face has integrated zero-egress storage with SkyPilot, allowing developers to run AI training workloads across multiple cloud providers (AWS, GCP, Azure) while storing models and data on Hugging Face without paying cloud egress fees. This simplifies moving AI projects between cloud platforms and cuts storage costs for independent developers and smaller teams.

Security researchers have discovered a new attack called 'BioShocking' that can trick AI browsers into bypassing their safety guardrails. The trick: present the AI with a game where incorrect answers (like 2+2=5) are rewarded. Once the AI accepts these 'alternate rules,' it enters a delusional state where normal restrictions no longer apply, allowing attackers to extract passwords, private code, and credentials. The attack worked against multiple AI browsers including ChatGPT Atlas, Comet, and Claude Chrome.

OpenAI CEO Sam Altman is reportedly discussing giving the US government a 5% stake in the company—worth roughly $320 per American household at current valuations. The proposal aims to address concerns that AI companies benefit from human work without compensation, and provide a safety net against AI-driven job losses. Separately, a leaked Treasury report compares the AI market to the dotcom bubble, contradicting the administration's public optimism. The details remain murky, but the proposal signals growing political pressure on AI companies to share wealth more broadly.

In May, an anonymous artist who goes by SHL0MS on X posted that he had used AI to generate an image inspired by Claude Monet and asked people to weigh in on how it missed the mark. More than 600 responses called out issues, saying the colors were off, the depth was all wrong, and that AI didn’t understand how light worked. SHL0MS then revealed that the image was of a real Monet, one of around 250 variations of water lilies the artist had painted in his lifetime. He had simply downloaded a high-resolution image from Wikimedia and cropped out the signature. He minted the exchange as an NFT (a unique digital collectible recording ownership of the work), titled it “Inferior Image,” and sold it for just over US $40,000 after 28 bids. The stunt exposed how charged the conversation around AI art

OpenAI has released global ChatGPT usage data showing the tool is becoming a daily habit for more people worldwide. Six months after signing up, users send 50% more messages daily and try twice as many features. Africa and Asia show the fastest growth rates, and non-English speakers now make up over half of active users. The top non-English languages are Spanish, Portuguese, and Arabic. The data suggests AI adoption is deepening — people aren't just experimenting, they're integrating it into regular workflows.

OpenAI has released GeneBench-Pro, a tough benchmark testing whether AI models can handle real scientific research judgment — deciding when data supports a conclusion, when to revise assumptions, and when findings are decision-ready. Human experts estimate each problem takes 20-40 hours and costs thousands of dollars. GPT-5.6 Sol currently solves about 32% of problems, up from under 5% for earlier models. Even at partial capability, AI could meaningfully accelerate scientific discovery by automating early-stage analysis currently done by expensive expert teams.

Google partnered with educators in Sierra Leone to test whether AI tutoring could help students learn math. Over 8 weeks, students using Google's Guided Learning tool made learning gains equivalent to 1.2 to 1.7 years of progress. The AI didn't just give answers — it used Socratic questioning, posing follow-up problems to help students understand concepts. Notably, 91% of student-AI conversations focused on building understanding rather than getting quick answers. Teachers also reported using AI to improve their own lesson planning.

A renowned computational linguist is setting the record straight on what her landmark 'stochastic parrots' paper actually meant—and what it didn't. Five years after warning that large language models repeat patterns without comprehension, Emily Bender says the term got stretched beyond its original scope. She argues that 'artificial intelligence' as a blanket term obscures more than it clarifies, grouping together vastly different technologies like chatbots and protein-folding tools under one hype-filled label.

The UK government is partnering with Google DeepMind to build an AI tool that helps planning officers process house-building applications faster. With the UK aiming to build 1.5 million homes by 2029, the new AI acts as a highly skilled assistant — cross-referencing policy documents, historical files, and PDFs that normally take hours to review manually. The goal is to cut decision times in half. Early trials are happening in three councils, with national rollout planned for 2027.

Google has released two new AI tools for creators: Nano Banana 2 Lite, which generates images in just 4 seconds at $0.034 per 1,000 images, and Gemini Omni Flash, which creates 10-second videos at $0.10 per second. The key feature is that developers can chain these together — generate an image fast with Nano Banana 2 Lite, then animate it into video with Gemini Omni Flash. Both models are available through Google's developer tools and will roll out to consumer products like Google Search, Photos, and Flow.

Australian Payments Plus (AP+), the operator of critical payment infrastructure used by millions daily, has rolled out ChatGPT Enterprise and Codex across its teams. The AI tools help employees handle complex payments rules, draft member communications, and run product simulations — reducing days of manual work to minutes in some cases. Nearly 80% of staff report being more creative or producing higher-quality work, and teams have built over 300 custom GPTs. The key lesson: giving employees secure, governed AI tools works better than top-down mandates.

Google DeepMind has launched a three-month accelerator program for 15 European robotics startups working on real-world AI applications. These companies span healthcare, manufacturing, construction, and ocean cleanup — from robots that perform brain surgery to ones that build houses. Each startup gets access to Google's AI models, technical mentorship, and robotics expertise to help turn their research into working products.

IBM Research has released ScarfBench, a new benchmark specifically testing how well AI agents handle enterprise Java framework migrations—like upgrading old codebases to modern frameworks. The benchmark evaluates AI agents on accuracy, efficiency, and error rates when tackling these complex, real-world software modernization tasks that many businesses face.

Hugging Face has launched 'Every Eval Ever' (EEE), a community-driven effort to aggregate and standardize AI model evaluation results directly on model pages. Instead of scattered benchmarks across different papers, users can now compare how models perform across standardized tests in one place.

India's Tata Consultancy Services partnered with French AI startup Mistral to become Mistral Forge's first global systems integrator partner. Forge lets enterprises build custom AI models using their own data for banking, healthcare, manufacturing, and public sector. TCS will set up a dedicated Mistral Centre of Excellence. This signals major IT services firms increasingly betting on European AI alternatives to OpenAI and Anthropic for enterprise deployments.

Singapore's data centres are getting hit by record 17% electricity price jumps, but most operators have fixed power contracts protecting them until 2027. When those expire, costs will likely be passed to tenants rather than causing operators to leave. The bigger shift: new AI data centre expansion may increasingly go to Johor and Batam instead of Singapore, where land and power are cheaper.

A new ILO study finds nearly 80 million ASEAN workers—about 23% of the workforce—hold jobs with some exposure to generative AI. Singapore leads the region with 42.2% of workers exposed, but researchers say large-scale job losses haven't materialized yet. The bigger concern is a 'preparedness gap': workers need skills to adapt, especially women who are twice as likely to be in high-exposure roles.

Data and AI company Databricks has appointed Raymond Wong as its Singapore Public Sector Lead, overseeing partnerships with government agencies including CPF Board, GovTech, Singapore Customs, and healthtech agency Synapxe. The appointment comes as Singapore refreshes its National AI Strategy to scale AI deployment across public services. Wong will focus on helping agencies unify data systems and make AI insights accessible through natural language queries.

Google Gemini's user base in Southeast Asia has more than doubled, with Singapore leading global per-capita adoption. The app's strength in local languages is key—nearly 70% of prompts across the region are in native languages. Google is also rolling out Gemini Spark, an AI agent that completes tasks for users, in the region this week.

DeepSeek founder Liang Wenfeng is now worth US$36 billion, making him the world's richest AI model creator and China's eighth-richest person. His net worth more than doubled after DeepSeek's massive US$7.4 billion funding round in June, valuing the company at US$50 billion. Unlike US peers who dilute heavily to investors, Liang retained roughly 78% of DeepSeek—a rare feat in modern AI.

Australian AI infrastructure firm Firmus is building a massive 360-megawatt data center campus in Batam, Indonesia—powered by Nvidia chips—with Singapore-based DayOne as a local partner. The facility, due in early 2027, will host up to 170,000 AI accelerator chips and serve AI companies rather than big cloud firms like AWS or Google. Nvidia's revenue-sharing deal is worth an estimated US$25-30 billion over six years.

Singapore's manufacturing sector surged 12.2% year-on-year in Q2/2026, powered by AI-driven demand for semiconductors and electronics. The broader economy expanded 5.7%, with the Ministry of Trade and Industry pointing to AI infrastructure buildout as the key driver. Electronics and precision engineering clusters led the charge, while chemicals struggled. For everyday Singaporeans, this growth signals continued demand for tech jobs and signals Singapore's role as a key player in the global AI supply chain.

Singapore is pushing to equip its workforce with AI skills through literacy programmes and training initiatives. The government-backed effort aims to help workers understand, use, and adapt to AI tools in their daily jobs. This comes as AI adoption grows across industries, making basic AI literacy increasingly important for career competitiveness.

SoftBank CEO Masayoshi Son predicts AI development will need $5 trillion yearly by 2040, calling bubble concerns "absurd." He's staking everything on AI with over $60 billion already in OpenAI, plus data centers and robotics. Son envisions 100 trillion AI agents making decisions independently by 2040, shifting from a human-centric to an "agent-centric" world.

Countries like the Philippines are racing to pass laws forcing companies to store data locally, hoping to attract big tech investments. But building these digital walls comes with risks—drones knocked out UAE and Bahrain data centers, disrupting services for weeks. Meanwhile, global chip shortages mean governments can pass laws today but wait years for the actual hardware to arrive.

Three in five Singapore PMETs can't tell AI-generated misinformation from real content, according to a poll of 1,150 professionals. Most don't bother checking original sources when AI summaries appear in search results or news. The National Library Board and KPMG have launched a year-long reading initiative to help workers develop critical thinking skills needed to spot AI fakes. Minister of State Rahayu Mahzam said companies should invest in human capabilities alongside AI adoption.

Research from Roland Berger and AGOS Asia shows 74% of Global Business Services leaders across Asia have an AI vision, but only 20% can prove it’s working. The gap is now a boardroom crisis—2025 AI budgets are being asked to show P&L returns by end of 2026. Finance functions face the hardest test: boards want cost reduction, working capital gains, and real-time fraud detection, not just time saved on invoices. The real blocker isn’t technology—it’s governance. Most organizations give AI data access without a tested incident response plan.

Origa is an AI concierge for high-ticket sales—real estate, luxury goods—that remembers everything across phone calls, chat, and WhatsApp, so buyers never repeat themselves and sales teams arrive fully briefed. The Dubai-founded startup raised $450,000 from Antler Singapore and counts Malaysian property developer S P Setia among clients, where it delivered an 828% return on investment. Enterprise contracts run $75,000-$150,000 annually.

Venture capital investors are getting more skeptical about AI company valuations as US$4.3 trillion in paper wealth remains trapped in private markets. Unlike past boom cycles, LPs now demand proof that startups can actually return cash—not just post high valuations. Smaller funds face a tough choice: overpay for hot AI companies or miss out on major opportunities.

Germany's media regulator has ruled that AI search tools like Google's AI Overviews and Perplexity are "content providers" under German law—not just platforms displaying others' content. This means AI companies can be held directly liable for inaccurate or harmful AI-generated information, just like traditional publishers. A Munich court previously found Google could be liable for false statements made by its AI Overview feature. Both companies can challenge the decisions through legal channels.

Apple is suing OpenAI, former Apple executive Tang Tan, and other former Apple employees for allegedly stealing trade secrets related to AI hardware. Apple is also suing io Products—an AI hardware startup co-founded by former Apple design chief Jony Ive that OpenAI acquired. Analyst Gene Munster says Apple's aggressive legal action signals it sees AI hardware (like AirPods with cameras) as crucial to its future beyond the iPhone. OpenAI denies stealing trade secrets.

AI startup Reflection has signed a computing infrastructure deal worth over $1 billion with cloud provider Nebius to secure the GPU resources needed to train and run large AI models. The deal highlights the intense competition among AI startups to lock up computing power, as GPU availability remains a bottleneck across the industry. For consumers, this competition for resources could influence which AI products eventually reach the market.

Australia will create an "Office of AI" within its Department of Prime Minister and Cabinet to coordinate AI regulation across all government ministries. Prime Minister Albanese calls this a "world-first" approach, comparing it to how government coordinated responses to civil aviation in the 1920s or genetics in the 1990s. Australia currently has no specific AI laws and relies on existing privacy and consumer protection rules plus a voluntary ethics framework. The new office aims to provide clearer rules for AI investment approvals and compliance.

Cybersecurity firm Check Point warns in its 2026 AI Security Report that AI has evolved beyond assisting hackers to actively running live cyberattacks without human guidance. This marks a dangerous shift in the threat landscape, where AI systems now autonomously probe networks, execute intrusions, and adapt in real time.

Samsung has started mass-producing its fastest-ever storage drive for AI data centers—the PM1763 SSD, designed for Nvidia's new Vera Rubin supercomputing platform launching later this year. The drive delivers more than double the speed of its predecessor, reaching up to 28,400 MB/s, and is built for liquid-cooled server environments that AI data centers increasingly use.

SK Telecom is partnering with Nvidia to build AI factory software for South Korea's national AI network project. The company will evaluate equipment from Samsung, HFR, Ericsson, and Nokia. This matters because it shows how traditional telecom operators are pivoting to become AI infrastructure providers, which could reshape how AI services get delivered across Asia.

The European Commission approved €659 million (US$751 million) in German state aid for four semiconductor factories in Germany. The funding, part of the EU Chips Act to reduce reliance on Asian and American chipmakers, will support first-of-a-kind facilities in the semiconductor supply chain.

LTM, an Indian IT services firm with 87,000 employees in 40 countries, has partnered with Anthropic to offer Claude AI tools to enterprise customers. LTM will embed Claude and its coding assistant Claude Code into its existing AI delivery platform, train employees through an 'AI1000' certification program, and set up a Claude Center of Excellence. The deal is part of Anthropic's $100 million partner program to expand enterprise AI adoption.

SINGAPORE – For the first time, artificial intelligence tools have automated [significant](https://www.straitstimes.com/world/ai-can-outpace-cybersecurity-norms-in-months-spy-alliance?ref=inline-article)[parts of cyber intrusions](https://www.straitstimes.com/world/ai-can-outpace-cybersecurity-norms-in-months-spy-alliance?ref=inline-article) over the past 12 months, according to a new report. In its 56-page report released on July 14, Check Point Research said that AI played a role in nearly every stage of a cyber attack process: social engineering, malware development, live intrusion, building attacker tools, and researching vulnerabilities. The intelligence arm of cybersecurity firm Check Point Software Technologies recorded instances where AI autonomously executed a cyberattacks over th

Huawei has launched its Pura 90s Series globally, featuring a 200MP telephoto camera, AI-powered photo editing tools like reflection removal and subject isolation, and its Celia AI assistant. The phones support up to 100W fast charging. For Singapore users, the lack of Google services remains a practical hurdle despite the impressive camera and AI features.

Workplace robots powered by AI could help autistic adults join the workforce by redesigning jobs around their needs - steadying work pace, catching errors instantly, and removing chaotic peripheral tasks. Currently, 84% of companies bolt robots onto old job designs instead of rebuilding roles around what the technology can now do.

Data analytics company Databricks is aggressively expanding its India operations, hiring hundreds of new staff and opening an R&D center in Bengaluru focused on AI and large language models. The company has nearly 1,000 employees in India and plans to deploy US$250 million over three years. Databricks, valued at US$134 billion, competes with Snowflake and counts Indian clients like Swiggy, Cred, and Freshworks. The move reflects a broader trend of tech companies shifting from support roles to deep-tech R&D work in India.

Google's Gemini app has doubled its user base in Southeast Asia, signaling growing consumer appetite for AI-powered tools in the region. The expansion reflects intensifying competition between major AI players like Google, OpenAI, and Meta for the lucrative SEA market.

Edge cloud platform Fastly has become the first in its sector to join DIMPACT, a coalition working to measure digital media's carbon footprint. With AI-generated internet traffic growing fast, companies want better data on their digital emissions to reduce environmental impact.

Sarvam AI, an Indian startup building large language models for Indian languages, is in talks to raise $320-350 million at a $1.5 billion valuation. Backers include Nvidia, Amazon, HCLTech, and Bessemer Venture Partners. The company was chosen by India's government to build the country's sovereign AI model and received thousands of subsidized Nvidia H100 chips. This reflects a global trend where governments back domestic AI companies to reduce reliance on foreign tech.

South Korea raised its 2026 economic growth forecast to 3.0%—its highest in five years—riding the wave of global AI chip demand. The government plans "mega projects" covering semiconductors, AI data centers, and physical AI, with next year's budget set to exceed $530 billion with 10% more spending. The finance ministry wants to push potential growth to 3% and raise per capita income to $50,000. High oil prices mean inflation is expected at 2.6% though.

Intel's stock has climbed 65% this year, boosted by new AI deals. Google will use Intel's Xeon 6 processors for AI training and inference. Intel also joined Terafab, a Texas project building custom chips for Elon Musk's companies including SpaceX, xAI, and Tesla. The chipmaker also bought back full control of its Ireland Fab 34 plant for $14.2 billion after previously selling a 49% stake.

China's exports jumped 27% in June compared to last year, far exceeding expectations, largely driven by surging demand for AI-related technology including semiconductors and electronics. A trade surplus of $125.6 billion was recorded. Exports to Southeast Asia surged nearly 35%, while shipments to the US rose almost 14%. Analysts say AI boom-driven semiconductor prices are the main driver, though China's export growth may become fragile depending on global demand and regulatory barriers like tariffs.

Slack has announced major AI integrations with Salesforce at Dreamforce 2025, allowing users to manage customer records, IT support, HR inquiries, and analytics dashboards directly within Slack. The updates include a rebuilt Slackbot and a Channel Expert AI agent that provides automated responses using company knowledge. Partners like OpenAI, Google, and Anthropic can now build AI agents that work inside Slack through new APIs. Enterprise teams should note that pricing and rollout timelines remain unclear, with some features in closed beta until early 2026.

Finland’s technology giant [Nokia](https://www.nokia.com/) has signed a 5G expansion agreement with Taiwan Mobile to accelerate the evolution toward AI-native mobile networks across Taiwan, deploying Nokia’s latest AirScale portfolio. In a statement on Tuesday, Nokia said the deployment would include next-generation baseband and radio solutions alongside AI-driven software capabilities. On network intelligence, Nokia will deploy AI-powered software for real-time automation and predictive analytics, including its Predictive Hardware Analytics service and MantaRay SON self-organizing network solution. Regarding infrastructure, next-generation baseband and advanced radio solutions will increase network capacity and uplink performance to meet demand from AI-generated traffic. For sustainabilit

Indian IT services firm LTM expects its AI business to eventually surpass its traditional services revenue, CEO Venu Lambu said. The company just signed a partnership with Anthropic to deploy Claude AI models to enterprise clients. LTM disclosed it currently earns $150 million quarterly from AI-native businesses, representing 12% of total revenue. Lambu acknowledged that clients are worried about token costs from AI usage-based pricing and said LTM helps businesses build governance frameworks to manage expenses.

HubSpot's founding CRO Mark Roberge warns AI startups that early revenue is a trap—high churn can hide behind initial sales hype. His advice: prove customers actually use the product daily and generate financial returns before hiring sales teams. Key red flags to avoid include hiring against fundraising milestones instead of operational performance, and mistaking market hype for real product-market fit. Boards should track retention and margin data, not just pipeline growth.

SoftBank has stalled talks to raise at least $6 billion through a margin loan backed by its OpenAI stake, after cutting the target from $10 billion. Banks are cautious about valuing shares in private companies like OpenAI. SoftBank has invested $2.2 billion in OpenAI through its Vision Fund 2 since 2024 and committed up to $30 billion in follow-on investments. OpenAI's confidential IPO filing valued the company at $852 billion.

Code Metal, an AI startup that translates code between programming languages while ensuring safety compliance, raised $36.5 million in Series A funding from Accel. The company targets defense, automotive, and semiconductor sectors where software errors can be dangerous. It claims it is on track to deliver eight figures in revenue this year, partly through Department of Defense software modernization contracts.

By Haripriya Suresh Mon, 13 July 2026 at 11:14 pm GMT-7 2 min read By Haripriya Suresh BENGALURU, July 14 (Reuters) - Indian software services firm LTM expects AI revenue to outpace its traditional services business, CEO Venu Lambu told ‌Reuters, betting that enterprises will need IT companies to deploy powerful large ‌language models from firms such as Anthropic and OpenAI. Investor worries that increasingly capable AI models could disrupt ​the businesses of traditional IT services firms have led to India's Nifty IT index falling more than 23% year-to-date, on course for its second-biggest loss since 2008. LTM, which on Monday signed a partnership with Anthropic to deploy Claude to enterprise clients, is betting that ‌these models will help create ⁠a new implementation market for IT firms

Singapore-based AI video startup PixVerse has secured an extended Series C round, pushing its valuation past $2 billion. The funding shows investors still see room for challengers in the crowded AI video generation space despite competition from OpenAI's Sora and other rivals. The company's latest V6 model produces more photorealistic video, and this valuation milestone signals growing confidence in AI video as a viable commercial market.

Research from Pangram Labs suggests nearly half of all LinkedIn posts are now AI-written. The finding highlights how deeply generative AI has渗透 professional self-presentation—people are outsourcing their LinkedIn voices to chatbots. The broader implications touch on authenticity, job searching, and how we signal professional identity online.

OpenAI fixed mysterious crashes in its ChatGPT data infrastructure (Rockset) that seemed impossible—functions returning to bogus memory addresses. Using an epidemiological approach (analyzing all crash data rather than individual cases), engineers discovered two separate bugs: a silent hardware failure on one Azure host where the CPU simply miscalculated, and an 18-year-old race condition in GNU libunwind library. The fix involved denylisting the bad host, improving crash detection, and changing VM reuse policies.

TechInAsia has compiled a list of 50 rising AI startups across Asia that have recently closed funding rounds. The list covers companies up to Series E stage and includes details on each funding round. This is premium content requiring a subscription to access the full details.

Google DeepMind CEO Demis Hassabis is calling for a global AI watchdog, with the US at the helm, that could pause AI development if frontier models become too dangerous. He proposes a FINRA-style body to set safety standards for the most advanced AI systems. This matters because as AI gets smarter, governments worldwide are scrambling to figure out who regulates what—and Hassabis, one of the industry's own, is saying the US should lead that conversation.

This is the same story as Article 0—DeepMind CEO Demis Hassabis's proposal for a global AI watchdog modeled after financial regulators. The key message: nobody really knows where AI is heading, so 'cautious optimism' means building safety guardrails now rather than waiting for a crisis.

Singapore-based AI video startup PixVerse has raised $439 million in a Series C extension, pushing its valuation past $2 billion. Founded in 2023 by former ByteDance computer vision engineers, PixVerse offers AI models that generate videos up to 4K resolution with audio. The company claims 150 million registered users and 15 million monthly active users on its consumer platform. It competes with ByteDance's Seedance, Runway, Midjourney, and others in the heated AI video generation market.

Uber's chief product officer outlined the company's strategy of expanding beyond rides and food delivery into travel services, financial products, and autonomous vehicles. The company is using AI to power features like voice-requested rides, an 'earner assistant' that suggests optimal driving locations, and grocery shopping by voice command. Uber also confirmed it sells data-labeling services to AI companies using its driver base. Despite adding hotels, boat rentals, and shopping features, Uber insists it's not trying to become an 'everything app' like Grab.

PsiQuantum, a quantum computing startup, is building a massive machine that uses particles of light (photons) to perform calculations that today's computers would take millions of years to solve. The company has raised $1 billion and is constructing facilities in Chicago and Australia, aiming to have working hardware by 2027. If successful, such computers could revolutionize drug discovery, materials science, and eventually AI by simulating quantum mechanics directly. The catch? No one has yet built a quantum computer powerful enough to be commercially useful—and experts remain skeptical about PsiQuantum's timeline.

JPMorgan CEO Jamie Dimon revealed that AI has helped the bank cut up to 40% of jobs in certain roles — the most specific workforce impact claim yet from a major bank leader. Combined with Bank of America and Goldman Sachs reporting strong results fueled by AI-driven market activity, the banking sector is showing concrete signs of AI replacing human work. Dimon acknowledged AI's transformative impact while remaining optimistic about the broader economy.

Apple is suing OpenAI for trade secrets, escalating legal battles for Sam Altman's company. The lawsuit centers on claims that OpenAI misused Apple's technology or information during their partnership discussions. With Apple having invested heavily in OpenAI and the company eyeing an IPO, this legal fight could reshape AI industry partnerships and how companies protect their competitive secrets. It's a high-stakes showdown between two tech giants with deep pockets and even deeper ambitions in AI.

OpenAI's GPT-5.6 Sol language model has solved the 50-year-old "cycle double cover conjecture" in mathematics — a problem about networks of dots and connections that had stumped human mathematicians for decades. The AI was given 64 sub-agents working in parallel and told not to give up for at least 8 hours. Mathematicians say this is another sign AI will fundamentally change how mathematical research is done.

New York has become the first US state to halt data center construction, signing a one-year pause on large facilities amid concerns over energy consumption and environmental impact. The governor also wants to end all tax incentives for data centers. While AI systems rely heavily on these massive computing facilities to function, communities have raised concerns about strain on power grids, water usage, and rising utility bills. This could set a precedent for other states and affect where tech companies choose to build their AI infrastructure.

New York has become the first US state to halt new data center construction. Governor Hochul signed an executive order pausing projects over 50 megawatts after public backlash over rising electricity bills, water shortages, and noise pollution. The one-year moratorium aims to establish environmental review rules before any new approvals. The move puts New York at odds with federal efforts to fast-track AI infrastructure development.

The biggest AI battleground may have shifted away from the most powerful closed models. Chinese open-weight AI models now account for 41% of downloads on Hugging Face, surpassing U.S. models. On some platforms, open-source models handle nearly a third of AI requests. Hugging Face CEO Clem Delangue argues that companies increasingly want to own and control their own AI rather than rely on expensive closed APIs. Microsoft CEO Satya Nadella recently warned that relying on a single AI provider creates dangerous lock-in. The debate pits those who say open models risk dangerous misuse against those who argue keeping AI in few hands is the real danger.

U.S. AI startup Reflection AI has signed a $1 billion deal with European infrastructure firm Nebius to access Nvidia chips. The company, valued at $8 billion and backed by Nvidia and Sequoia, is building open AI models to compete with top closed-source providers. This comes weeks after a similar SpaceX compute deal. Nebius itself has major deals with Meta worth up to $27 billion.

The Linux Foundation has launched the x402 Foundation, backed by major players like Coinbase, AWS, Google, Visa, and Mastercard, to create a standard payment protocol for AI agents. The goal is to let AI bots pay for things online without friction, fees, or waiting. It aims to enable AI agents to autonomously buy services, APIs, and resources as they work.

Google is adding AI image generation directly into Search. When no matching images exist on the web for your query, its Nano Banana 2 Lite model will create one on the fly. The feature rolls out to users in the coming weeks, making Search more visual and AI-powered than ever before.

Waze just got smarter with AI. The navigation app now lets you report road issues by simply talking, get routes that learn your preferences (highway vs local roads), and even find places by asking conversational questions like 'Find me cheap gas nearby.' Motorcyclists in Malaysia, Philippines, and several Latin American countries get special two-wheeler routing that warns about potholes and narrow bridges. A new 'less chatty' mode cuts down voice prompts for drivers who want to focus on music or podcasts.

Meta announced a $50 billion Louisiana data center while simultaneously planning to sell excess AI computing power to others—raising questions about whether the AI industry is overbuilding. Investors cheered the move, but chipmakers like Micron dropped 10% and AMD fell 7%. The market is starting to question whether the massive AI infrastructure spending will actually pay off. Major tech firms plan to spend $725 billion on AI data centers in 2026 alone.

Sarvam AI, an Indian startup building large language models for Indian languages, is in talks to raise $320-350 million at a $1.5 billion valuation. Backers include Nvidia, Amazon, HCLTech, and Bessemer Venture Partners. The company was chosen by India's government to build the country's sovereign AI model and received thousands of subsidized Nvidia H100 chips. This reflects a global trend where governments back domestic AI companies to reduce reliance on foreign tech.

TechInAsia has compiled a list of 50 rising AI startups across Asia that have recently closed funding rounds. The list covers companies up to Series E stage and includes details on each funding round. This is premium content requiring a subscription to access the full details.

Singapore-based AI video startup PixVerse has raised $439 million in a Series C extension, pushing its valuation past $2 billion. Founded in 2023 by former ByteDance computer vision engineers, PixVerse offers AI models that generate videos up to 4K resolution with audio. The company claims 150 million registered users and 15 million monthly active users on its consumer platform. It competes with ByteDance's Seedance, Runway, Midjourney, and others in the heated AI video generation market.

Countries like the Philippines are racing to pass laws forcing companies to store data locally, hoping to attract big tech investments. But building these digital walls comes with risks—drones knocked out UAE and Bahrain data centers, disrupting services for weeks. Meanwhile, global chip shortages mean governments can pass laws today but wait years for the actual hardware to arrive.

Intel's stock has climbed 65% this year, boosted by new AI deals. Google will use Intel's Xeon 6 processors for AI training and inference. Intel also joined Terafab, a Texas project building custom chips for Elon Musk's companies including SpaceX, xAI, and Tesla. The chipmaker also bought back full control of its Ireland Fab 34 plant for $14.2 billion after previously selling a 49% stake.

SoftBank has stalled talks to raise at least $6 billion through a margin loan backed by its OpenAI stake, after cutting the target from $10 billion. Banks are cautious about valuing shares in private companies like OpenAI. SoftBank has invested $2.2 billion in OpenAI through its Vision Fund 2 since 2024 and committed up to $30 billion in follow-on investments. OpenAI's confidential IPO filing valued the company at $852 billion.

Code Metal, an AI startup that translates code between programming languages while ensuring safety compliance, raised $36.5 million in Series A funding from Accel. The company targets defense, automotive, and semiconductor sectors where software errors can be dangerous. It claims it is on track to deliver eight figures in revenue this year, partly through Department of Defense software modernization contracts.

Uber's chief product officer outlined the company's strategy of expanding beyond rides and food delivery into travel services, financial products, and autonomous vehicles. The company is using AI to power features like voice-requested rides, an 'earner assistant' that suggests optimal driving locations, and grocery shopping by voice command. Uber also confirmed it sells data-labeling services to AI companies using its driver base. Despite adding hotels, boat rentals, and shopping features, Uber insists it's not trying to become an 'everything app' like Grab.

Research from Roland Berger and AGOS Asia shows 74% of Global Business Services leaders across Asia have an AI vision, but only 20% can prove it’s working. The gap is now a boardroom crisis—2025 AI budgets are being asked to show P&L returns by end of 2026. Finance functions face the hardest test: boards want cost reduction, working capital gains, and real-time fraud detection, not just time saved on invoices. The real blocker isn’t technology—it’s governance. Most organizations give AI data access without a tested incident response plan.

Research from Pangram Labs suggests nearly half of all LinkedIn posts are now AI-written. The finding highlights how deeply generative AI has渗透 professional self-presentation—people are outsourcing their LinkedIn voices to chatbots. The broader implications touch on authenticity, job searching, and how we signal professional identity online.

Origa is an AI concierge for high-ticket sales—real estate, luxury goods—that remembers everything across phone calls, chat, and WhatsApp, so buyers never repeat themselves and sales teams arrive fully briefed. The Dubai-founded startup raised $450,000 from Antler Singapore and counts Malaysian property developer S P Setia among clients, where it delivered an 828% return on investment. Enterprise contracts run $75,000-$150,000 annually.

South Korea raised its 2026 economic growth forecast to 3.0%—its highest in five years—riding the wave of global AI chip demand. The government plans "mega projects" covering semiconductors, AI data centers, and physical AI, with next year's budget set to exceed $530 billion with 10% more spending. The finance ministry wants to push potential growth to 3% and raise per capita income to $50,000. High oil prices mean inflation is expected at 2.6% though.

Singapore's manufacturing sector surged 12.2% year-on-year in Q2/2026, powered by AI-driven demand for semiconductors and electronics. The broader economy expanded 5.7%, with the Ministry of Trade and Industry pointing to AI infrastructure buildout as the key driver. Electronics and precision engineering clusters led the charge, while chemicals struggled. For everyday Singaporeans, this growth signals continued demand for tech jobs and signals Singapore's role as a key player in the global AI supply chain.

Singapore is pushing to equip its workforce with AI skills through literacy programmes and training initiatives. The government-backed effort aims to help workers understand, use, and adapt to AI tools in their daily jobs. This comes as AI adoption grows across industries, making basic AI literacy increasingly important for career competitiveness.

China's exports jumped 27% in June compared to last year, far exceeding expectations, largely driven by surging demand for AI-related technology including semiconductors and electronics. A trade surplus of $125.6 billion was recorded. Exports to Southeast Asia surged nearly 35%, while shipments to the US rose almost 14%. Analysts say AI boom-driven semiconductor prices are the main driver, though China's export growth may become fragile depending on global demand and regulatory barriers like tariffs.

Data analytics company Databricks is aggressively expanding its India operations, hiring hundreds of new staff and opening an R&D center in Bengaluru focused on AI and large language models. The company has nearly 1,000 employees in India and plans to deploy US$250 million over three years. Databricks, valued at US$134 billion, competes with Snowflake and counts Indian clients like Swiggy, Cred, and Freshworks. The move reflects a broader trend of tech companies shifting from support roles to deep-tech R&D work in India.

Slack has announced major AI integrations with Salesforce at Dreamforce 2025, allowing users to manage customer records, IT support, HR inquiries, and analytics dashboards directly within Slack. The updates include a rebuilt Slackbot and a Channel Expert AI agent that provides automated responses using company knowledge. Partners like OpenAI, Google, and Anthropic can now build AI agents that work inside Slack through new APIs. Enterprise teams should note that pricing and rollout timelines remain unclear, with some features in closed beta until early 2026.

SoftBank CEO Masayoshi Son predicts AI development will need $5 trillion yearly by 2040, calling bubble concerns "absurd." He's staking everything on AI with over $60 billion already in OpenAI, plus data centers and robotics. Son envisions 100 trillion AI agents making decisions independently by 2040, shifting from a human-centric to an "agent-centric" world.

Google's Gemini app has doubled its user base in Southeast Asia, signaling growing consumer appetite for AI-powered tools in the region. The expansion reflects intensifying competition between major AI players like Google, OpenAI, and Meta for the lucrative SEA market.

HubSpot's founding CRO Mark Roberge warns AI startups that early revenue is a trap—high churn can hide behind initial sales hype. His advice: prove customers actually use the product daily and generate financial returns before hiring sales teams. Key red flags to avoid include hiring against fundraising milestones instead of operational performance, and mistaking market hype for real product-market fit. Boards should track retention and margin data, not just pipeline growth.

OpenAI fixed mysterious crashes in its ChatGPT data infrastructure (Rockset) that seemed impossible—functions returning to bogus memory addresses. Using an epidemiological approach (analyzing all crash data rather than individual cases), engineers discovered two separate bugs: a silent hardware failure on one Azure host where the CPU simply miscalculated, and an 18-year-old race condition in GNU libunwind library. The fix involved denylisting the bad host, improving crash detection, and changing VM reuse policies.

Three in five Singapore PMETs can't tell AI-generated misinformation from real content, according to a poll of 1,150 professionals. Most don't bother checking original sources when AI summaries appear in search results or news. The National Library Board and KPMG have launched a year-long reading initiative to help workers develop critical thinking skills needed to spot AI fakes. Minister of State Rahayu Mahzam said companies should invest in human capabilities alongside AI adoption.

Finland’s technology giant [Nokia](https://www.nokia.com/) has signed a 5G expansion agreement with Taiwan Mobile to accelerate the evolution toward AI-native mobile networks across Taiwan, deploying Nokia’s latest AirScale portfolio. In a statement on Tuesday, Nokia said the deployment would include next-generation baseband and radio solutions alongside AI-driven software capabilities. On network intelligence, Nokia will deploy AI-powered software for real-time automation and predictive analytics, including its Predictive Hardware Analytics service and MantaRay SON self-organizing network solution. Regarding infrastructure, next-generation baseband and advanced radio solutions will increase network capacity and uplink performance to meet demand from AI-generated traffic. For sustainabilit

Indian IT services firm LTM expects its AI business to eventually surpass its traditional services revenue, CEO Venu Lambu said. The company just signed a partnership with Anthropic to deploy Claude AI models to enterprise clients. LTM disclosed it currently earns $150 million quarterly from AI-native businesses, representing 12% of total revenue. Lambu acknowledged that clients are worried about token costs from AI usage-based pricing and said LTM helps businesses build governance frameworks to manage expenses.

By Haripriya Suresh Mon, 13 July 2026 at 11:14 pm GMT-7 2 min read By Haripriya Suresh BENGALURU, July 14 (Reuters) - Indian software services firm LTM expects AI revenue to outpace its traditional services business, CEO Venu Lambu told ‌Reuters, betting that enterprises will need IT companies to deploy powerful large ‌language models from firms such as Anthropic and OpenAI. Investor worries that increasingly capable AI models could disrupt ​the businesses of traditional IT services firms have led to India's Nifty IT index falling more than 23% year-to-date, on course for its second-biggest loss since 2008. LTM, which on Monday signed a partnership with Anthropic to deploy Claude to enterprise clients, is betting that ‌these models will help create ⁠a new implementation market for IT firms

Ransomware attacks on small and medium businesses across Southeast Asia climbed in early 2026. Singapore saw SMB targeting rise from 0.57% to 0.69%. Security firm Kaspersky warns that SMBs are increasingly seen as easy entry points into larger supply chains, and most attackers now use double extortion tactics.

Twenty-six Meta employees are suing the tech giant, alleging it used AI systems to pick people for layoffs in ways that unfairly hit workers who took medical, parental, or family leave. The lawsuit, filed in California federal court, claims Meta's AI tracked employee activity and keystrokes without adjusting scores for legally protected leave—meaning workers who took pregnancy or medical leave got penalized simply for being away. About half the plaintiffs took pregnancy or caregiving leave. Meta denies the claims, saying all decisions were made by humans. The case could set a major precedent for how companies must audit AI tools used in hiring and firing.

OpenAI's Greg Brockman says the next AI shift is moving from answering questions to doing actual work. He envisions AI agents replacing app screens, where users talk to a persistent AI that handles tasks across software. But this only works if trust systems are solid — companies need clear controls over what agents can see, draft, approve, or execute. Brockman also argues compute scarcity will keep AI expensive even as per-task costs fall, and that expert knowledge in fields like healthcare will matter more than raw general intelligence.

The White House announced the creation of a new AI and cybersecurity coordination group to strengthen cooperation between government agencies on technology security issues. The group aims to streamline how the US government handles AI-related threats and policy decisions. This comes as AI adoption accelerates across industries, raising concerns about data security and national competitiveness.

UK banks cannot access Anthropic's powerful Mythos AI model, which is crucial for spotting cybersecurity weaknesses. A government-appointed AI adviser is calling for Britain to build its own AI capabilities instead of relying on US tech firms. The UK government has accepted recommendations to boost AI adoption in financial services.

A senior US Commerce official confirmed that regulations on AI and semiconductors are on the way. The Trump administration will not replace the previous Biden-era rule that caps AI chip shipments to certain countries. This signals tighter control over AI technology exports globally.

Logitech has launched two new AI-powered conference cameras designed to make hybrid meetings smoother. The Rally AI Camera and Rally AI Camera Pro automatically frame participants—whether a group, individual speaker, or grid layout—using AI that mimics professional filmmaking techniques. The Pro model adds 15x zoom for large spaces. Both also track room occupancy to help companies manage office space more efficiently. For anyone who sits through daily video calls, this signals cameras are getting smarter at handling the awkward logistics of remote participation.

SK Telecom is partnering with Nvidia to build AI factory software for South Korea's national AI network project. The company will evaluate equipment from Samsung, HFR, Ericsson, and Nokia. This matters because it shows how traditional telecom operators are pivoting to become AI infrastructure providers, which could reshape how AI services get delivered across Asia.

Huawei has launched its Pura 90s Series globally, featuring a 200MP telephoto camera, AI-powered photo editing tools like reflection removal and subject isolation, and its Celia AI assistant. The phones support up to 100W fast charging. For Singapore users, the lack of Google services remains a practical hurdle despite the impressive camera and AI features.

A new ILO study finds nearly 80 million ASEAN workers—about 23% of the workforce—hold jobs with some exposure to generative AI. Singapore leads the region with 42.2% of workers exposed, but researchers say large-scale job losses haven't materialized yet. The bigger concern is a 'preparedness gap': workers need skills to adapt, especially women who are twice as likely to be in high-exposure roles.

Data and AI company Databricks has appointed Raymond Wong as its Singapore Public Sector Lead, overseeing partnerships with government agencies including CPF Board, GovTech, Singapore Customs, and healthtech agency Synapxe. The appointment comes as Singapore refreshes its National AI Strategy to scale AI deployment across public services. Wong will focus on helping agencies unify data systems and make AI insights accessible through natural language queries.

DeepSeek founder Liang Wenfeng is now worth US$36 billion, making him the world's richest AI model creator and China's eighth-richest person. His net worth more than doubled after DeepSeek's massive US$7.4 billion funding round in June, valuing the company at US$50 billion. Unlike US peers who dilute heavily to investors, Liang retained roughly 78% of DeepSeek—a rare feat in modern AI.

Venture capital investors are getting more skeptical about AI company valuations as US$4.3 trillion in paper wealth remains trapped in private markets. Unlike past boom cycles, LPs now demand proof that startups can actually return cash—not just post high valuations. Smaller funds face a tough choice: overpay for hot AI companies or miss out on major opportunities.

Cybersecurity firm Check Point warns in its 2026 AI Security Report that AI has evolved beyond assisting hackers to actively running live cyberattacks without human guidance. This marks a dangerous shift in the threat landscape, where AI systems now autonomously probe networks, execute intrusions, and adapt in real time.

Samsung has started mass-producing its fastest-ever storage drive for AI data centers—the PM1763 SSD, designed for Nvidia's new Vera Rubin supercomputing platform launching later this year. The drive delivers more than double the speed of its predecessor, reaching up to 28,400 MB/s, and is built for liquid-cooled server environments that AI data centers increasingly use.

Google Gemini's user base in Southeast Asia has more than doubled, with Singapore leading global per-capita adoption. The app's strength in local languages is key—nearly 70% of prompts across the region are in native languages. Google is also rolling out Gemini Spark, an AI agent that completes tasks for users, in the region this week.

Australian AI infrastructure firm Firmus is building a massive 360-megawatt data center campus in Batam, Indonesia—powered by Nvidia chips—with Singapore-based DayOne as a local partner. The facility, due in early 2027, will host up to 170,000 AI accelerator chips and serve AI companies rather than big cloud firms like AWS or Google. Nvidia's revenue-sharing deal is worth an estimated US$25-30 billion over six years.

The European Commission approved €659 million (US$751 million) in German state aid for four semiconductor factories in Germany. The funding, part of the EU Chips Act to reduce reliance on Asian and American chipmakers, will support first-of-a-kind facilities in the semiconductor supply chain.

LTM, an Indian IT services firm with 87,000 employees in 40 countries, has partnered with Anthropic to offer Claude AI tools to enterprise customers. LTM will embed Claude and its coding assistant Claude Code into its existing AI delivery platform, train employees through an 'AI1000' certification program, and set up a Claude Center of Excellence. The deal is part of Anthropic's $100 million partner program to expand enterprise AI adoption.

AI startup Reflection has signed a computing infrastructure deal worth over $1 billion with cloud provider Nebius to secure the GPU resources needed to train and run large AI models. The deal highlights the intense competition among AI startups to lock up computing power, as GPU availability remains a bottleneck across the industry. For consumers, this competition for resources could influence which AI products eventually reach the market.

Google DeepMind CEO Demis Hassabis is calling for a global AI watchdog, with the US at the helm, that could pause AI development if frontier models become too dangerous. He proposes a FINRA-style body to set safety standards for the most advanced AI systems. This matters because as AI gets smarter, governments worldwide are scrambling to figure out who regulates what—and Hassabis, one of the industry's own, is saying the US should lead that conversation.

This is the same story as Article 0—DeepMind CEO Demis Hassabis's proposal for a global AI watchdog modeled after financial regulators. The key message: nobody really knows where AI is heading, so 'cautious optimism' means building safety guardrails now rather than waiting for a crisis.

Germany's media regulator has ruled that AI search tools like Google's AI Overviews and Perplexity are "content providers" under German law—not just platforms displaying others' content. This means AI companies can be held directly liable for inaccurate or harmful AI-generated information, just like traditional publishers. A Munich court previously found Google could be liable for false statements made by its AI Overview feature. Both companies can challenge the decisions through legal channels.

Apple is suing OpenAI, former Apple executive Tang Tan, and other former Apple employees for allegedly stealing trade secrets related to AI hardware. Apple is also suing io Products—an AI hardware startup co-founded by former Apple design chief Jony Ive that OpenAI acquired. Analyst Gene Munster says Apple's aggressive legal action signals it sees AI hardware (like AirPods with cameras) as crucial to its future beyond the iPhone. OpenAI denies stealing trade secrets.

Australia will create an "Office of AI" within its Department of Prime Minister and Cabinet to coordinate AI regulation across all government ministries. Prime Minister Albanese calls this a "world-first" approach, comparing it to how government coordinated responses to civil aviation in the 1920s or genetics in the 1990s. Australia currently has no specific AI laws and relies on existing privacy and consumer protection rules plus a voluntary ethics framework. The new office aims to provide clearer rules for AI investment approvals and compliance.

Singapore-based AI video startup PixVerse has secured an extended Series C round, pushing its valuation past $2 billion. The funding shows investors still see room for challengers in the crowded AI video generation space despite competition from OpenAI's Sora and other rivals. The company's latest V6 model produces more photorealistic video, and this valuation milestone signals growing confidence in AI video as a viable commercial market.

SINGAPORE – For the first time, artificial intelligence tools have automated [significant](https://www.straitstimes.com/world/ai-can-outpace-cybersecurity-norms-in-months-spy-alliance?ref=inline-article)[parts of cyber intrusions](https://www.straitstimes.com/world/ai-can-outpace-cybersecurity-norms-in-months-spy-alliance?ref=inline-article) over the past 12 months, according to a new report. In its 56-page report released on July 14, Check Point Research said that AI played a role in nearly every stage of a cyber attack process: social engineering, malware development, live intrusion, building attacker tools, and researching vulnerabilities. The intelligence arm of cybersecurity firm Check Point Software Technologies recorded instances where AI autonomously executed a cyberattacks over th

Workplace robots powered by AI could help autistic adults join the workforce by redesigning jobs around their needs - steadying work pace, catching errors instantly, and removing chaotic peripheral tasks. Currently, 84% of companies bolt robots onto old job designs instead of rebuilding roles around what the technology can now do.

Edge cloud platform Fastly has become the first in its sector to join DIMPACT, a coalition working to measure digital media's carbon footprint. With AI-generated internet traffic growing fast, companies want better data on their digital emissions to reduce environmental impact.

Ransomware attacks on small and medium businesses across Southeast Asia climbed in early 2026. Singapore saw SMB targeting rise from 0.57% to 0.69%. Security firm Kaspersky warns that SMBs are increasingly seen as easy entry points into larger supply chains, and most attackers now use double extortion tactics.

UK banks cannot access Anthropic's powerful Mythos AI model, which is crucial for spotting cybersecurity weaknesses. A government-appointed AI adviser is calling for Britain to build its own AI capabilities instead of relying on US tech firms. The UK government has accepted recommendations to boost AI adoption in financial services.

A senior US Commerce official confirmed that regulations on AI and semiconductors are on the way. The Trump administration will not replace the previous Biden-era rule that caps AI chip shipments to certain countries. This signals tighter control over AI technology exports globally.

The White House announced the creation of a new AI and cybersecurity coordination group to strengthen cooperation between government agencies on technology security issues. The group aims to streamline how the US government handles AI-related threats and policy decisions. This comes as AI adoption accelerates across industries, raising concerns about data security and national competitiveness.

Logitech has launched two new AI-powered conference cameras designed to make hybrid meetings smoother. The Rally AI Camera and Rally AI Camera Pro automatically frame participants—whether a group, individual speaker, or grid layout—using AI that mimics professional filmmaking techniques. The Pro model adds 15x zoom for large spaces. Both also track room occupancy to help companies manage office space more efficiently. For anyone who sits through daily video calls, this signals cameras are getting smarter at handling the awkward logistics of remote participation.

Twenty-six Meta employees are suing the tech giant, alleging it used AI systems to pick people for layoffs in ways that unfairly hit workers who took medical, parental, or family leave. The lawsuit, filed in California federal court, claims Meta's AI tracked employee activity and keystrokes without adjusting scores for legally protected leave—meaning workers who took pregnancy or medical leave got penalized simply for being away. About half the plaintiffs took pregnancy or caregiving leave. Meta denies the claims, saying all decisions were made by humans. The case could set a major precedent for how companies must audit AI tools used in hiring and firing.