Singapore AI News & Daily Briefing

Singapore is expanding free AI and digital tools to help small and medium retail businesses compete online. The support includes training, subsidies, and ready-made AI solutions for tasks like inventory management, customer service, and marketing. This aims to help local shops digitalize without needing deep tech expertise.

Taiwanese prosecutors have detained two employees of AI server maker Super Micro Computer as part of an investigation into alleged illegal exports of Nvidia AI chips to China. Four employees total are being investigated for falsifying documents. This follows raids on Super Micro's Taiwan offices earlier this week, and prosecutors are now considering criminalising such exports altogether to give themselves more legal tools to stop the smuggling.

Nvidia is launching a revenue-sharing model that lets AI startups access its powerful computing resources without massive upfront costs. Instead of buying expensive AI hardware, smaller companies can pay through a cut of their future sales. Nvidia acts as a middleman connecting data centre operators with cloud providers, creating a new recurring income stream for itself while helping researchers and cash-strapped startups get into the AI race faster.

Google DeepMind and partners including Schmidt Sciences and ARIA are making $10 million available for researchers worldwide to study multi-agent AI safety. The funding call targets how millions of AI agents built by different companies will interact, communicate, and potentially cause unexpected collective problems. Researchers can apply in areas like test environments, agent network behavior, infrastructure security, and monitoring systems.

OpenAI's own data shows AI agents like Codex are fundamentally changing how work gets done. Non-developer adoption grew 137x in under a year—faster than developer adoption. Legal, recruiting, and finance teams now use Codex as their primary AI tool, generating 99.8% of the company's AI output tokens. Workers increasingly hand off 8-hour tasks to agents, with some users running 60 hours of agent work in a single day.

Google DeepMind has released a detailed security framework called the 'AI Control Roadmap' to keep AI agents safe as they become more powerful. The system treats potentially misaligned agents like insider threats—monitoring their actions, blocking harmful ones before they happen, and scaling defenses as AI capabilities grow. Google has already tested this on a million coding agent tasks, finding most problems stem from agents being overeager rather than malicious.

Google's new Gemma 4 12B is a compact AI model that can run on regular laptops with 16GB of RAM—no cloud needed. Unlike typical multimodal models, it processes images and audio directly without separate encoder systems, making it faster and more memory-efficient. With over 150 million downloads of the Gemma family, this open-source model under Apache 2.0 license brings advanced AI capabilities to everyday developers and consumers.

Capital One hired a former Amazon Alexa AI leader as Chief Scientist to build an internal research organization focused on domain-specific AI challenges in banking. The bank has launched what it claims is the first fully in-house agentic AI customer service tool built by a bank, and leads in AI patents among financial institutions.

Space-based AI data centers are nowhere close to reality, despite the hype. Experts calculate that Musk's plan would need 16,666 Starship launches and 25 years just to manufacture enough satellites. Cooling a single GPU in space requires 1.4 square meters of radiator, making the technology impractical for years, if ever.

Notion is shutting down its email app Notion Mail on September 22, saying most users now prefer AI agents to handle email instead. Users who relied on the Gmail client will need to export drafts before the cutoff date. The move marks the end of what remained of Skiff, an encrypted email service Notion acquired in 2024.

AI Singapore has new leadership in place, and the organisation is doubling down on its national AI ambitions. The piece explores how fresh direction is shaping the country's push to become a regional AI hub through research, talent development, and enterprise adoption.

A feature exploring how sports and media organizations are using AI, data, and technology to win over fans in Asia-Pacific, particularly around major events like the World Cup. It covers how AI helps personalize fan experiences, target advertising, and analyze engagement data. For sports fans in Singapore and the region, this means more tailored content and interactive experiences when watching major tournaments.

Meta is rolling out Muse Image, its new AI image generator, across Instagram, WhatsApp, and its advertising platform. Users can create and edit images directly in these apps using text prompts, while advertisers get new AI-powered creative tools. All AI-generated images will include watermarks for transparency.

Singapore desk workers are among the least sceptical about AI globally—only 29% identify as sceptics versus 37% worldwide—yet only 6% actually use AI daily in their work, nearly half the global average. The main complaints are generic outputs, lack of trust, and tools that don't fit their business needs. Researchers say Singapore workers aren't resisting AI; they're waiting for tools that actually work for them.

Singapore is pushing for AI adoption that benefits everyone — not just tech companies or big businesses. This means ensuring workers at all levels, including older employees and smaller firms, can use and benefit from AI tools. The approach focuses on inclusive growth so no one gets left behind as AI reshapes the economy.

Wealthy AI workers in San Francisco are driving house prices to record highs, with the median home now costing $1.76 million—up 19% from last year. Some sellers are even accepting shares in AI companies like OpenAI or Anthropic instead of cash. OpenAI employees alone have sold over $6 billion in shares recently, creating a new class of ultra-rich tech workers who are reshaping the city's property market.

Temasek-backed STT GDC and Nasdaq-listed SuperX AI Technology have opened an AI Innovation Center at STT Singapore 5 in Tai Seng. The facility lets enterprises test and run AI workloads locally using Nvidia Blackwell GPUs, Intel Xeon Gold processors, and 400G InfiniBand networking, with a 14-day free trial for qualified users. A survey found most Asian enterprise leaders are stuck in AI adoption due to lacking infrastructure.

China's Lingyi iTech, which makes smartphone and laptop components for Apple, Samsung, and Huawei, has raised approximately US$1.1 billion through its Hong Kong IPO. About 37.6% of proceeds—roughly US$391 million—will fund expansion into AI servers, humanoid robot hardware, and AI optical communication infrastructure. The company is allocating US$218 million over three years specifically for high-density AI servers and robotics manufacturing. This signals a major Chinese supplier pivoting from traditional consumer electronics components toward AI-era hardware.

Singapore-based Carousell turned its first full-year EBITDA profit in 2025, posting US$141 million in revenue (up 18%) driven by a 40% surge in recommerce sales. The classifieds giant is betting big on AI to automate listings, suggest prices, and crack down on scams—aiming to weave AI into every step of buying and selling. It also opened 29 physical trade-in stores by end-2025 and is eyeing more this year.

1MX AI, a Singapore-based venture capital firm founded in 2025, is raising US$15 million for its first fund. The firm has already invested in 16 AI startups—including a US humanoid robotics startup and a French quantum computing firm—using capital from its five founders and high-net-worth individuals. Unlike typical VC "spray and pray" strategies, 1MX takes a high-conviction, selective approach with ticket sizes between US$50,000 and US$500,000. The founders come from backgrounds in German acceleration, Asian M&A, Australian construction tech, and investment platforms.

Physical AI—which embeds advanced AI models into robots, autonomous vehicles, and manufacturing equipment—is creating new cyber-physical risks that corporate boards must address. Unlike traditional cybersecurity where breaches lead to data theft, Physical AI breaches can cause physical damage, supply chain disruptions, or injuries. The article cites the 2026 Anthropic Claude Mythos breach as an example of AI models autonomously exploiting vulnerabilities at machine speed. Boards need to shift from backward-looking compliance checklists to real-time resilience strategies.

Chinese AI firms MiniMax and Zhipu have seen their stock prices more than double since their January 2026 IPOs, with Zhipu shares rising 524%. Investors are bullish on domestic AI champions despite these companies spending roughly $10 for every $1 in revenue. Analysts see potential for growth but warn sustained earnings are needed to justify valuations.

Israeli cloud security startup Upwind has raised $250 million in Series B funding, reaching unicorn status at $1.5 billion valuation. The company reported 900% revenue growth and doubled its customer base since late 2024, serving clients like Siemens, Roku, and NuBank.

US automation firm BigPanda has acquired Israeli AI startup Velocity, which built an AI system that learns from past IT incidents to speed up troubleshooting. Velocity's founder will join BigPanda as VP of AI detection. The deal aims to reduce manual work in enterprise IT operations by combining AI-driven incident response.

This opinion piece argues that organizations need to move beyond simple data residency compliance to achieve true data sovereignty in the AI era. It suggests identifying sensitive data and critical workloads, then building sovereignty capabilities gradually. The piece emphasizes that full control over data is becoming a competitive necessity as AI adoption grows.

NCS, the technology arm of Singapore telco Singtel, is expanding its enterprise AI offerings across Asia-Pacific. The move signals how major Singapore tech firms are pushing AI tools to regional businesses for automation, customer service, and data analysis. For everyday workers, this could mean more AI-powered services from companies they interact with, from smarter chatbots to automated workflows.

Victory Giant Technology, a Chinese printed circuit board maker that supplies AI server components to Nvidia, launched a $2.2 billion Hong Kong IPO. The company saw quarterly profits surge 339% year-on-year, with over 70% of its business tied to Nvidia orders. Most proceeds will fund production expansion in mainland China and Southeast Asia, including Thailand and Vietnam.

Moneta, a Singapore-based AI startup, has developed an AI credit underwriter that helps banks process small business loan applications in minutes instead of weeks. The platform analyzes documents, generates credit memos, and has helped clients cut costs by 83%. The startup, which works primarily with Indonesian banks, recently raised $645,000 in pre-seed funding and serves 55 customers.

Insurance company Allianz Partners announced plans to cut up to 1,800 jobs, with the company citing AI-driven changes as a key reason for the workforce reduction. This follows a broader trend of companies using automation to streamline operations.

Singapore is building AI models trained specifically on local patient data to improve diagnosis of diabetes, high cholesterol, and eye diseases under the SIMFONI initiative. Health Minister Ong Ye Kung announced the program at the NCS Impact 2026 conference, highlighting that most current healthcare AI is trained on Western populations and may not work as well for Singaporeans. For example, Asians tend to develop diabetes at lower body weight than Westerners. The AI will suggest treatment options based on Singapore's clinical guidelines, but doctors will make the final decisions.

French AI company Mistral AI has raised €2 billion (US$2 billion) in a Series C funding round, valuing the company at €13.8 billion. Major investors include ASML (which invested €1.3 billion), Nvidia, and Andreessen Horowitz. The funding will fuel AI research for industrial and engineering sectors. Notably, ASML's investment aims to integrate AI into its chip-making lithography machines. Mistral, founded in 2023, now has over 350 employees and maintains an office in Singapore alongside Paris, London, Palo Alto, and New York.

HSBC has signed a multi-year deal with French AI company Mistral AI to deploy self-hosted AI models across the global bank. The models will run on HSBC's internal infrastructure for tasks like multilingual translation, financial analysis, and risk identification. This approach keeps sensitive customer data within the bank's own systems rather than relying on external AI providers, a key concern for financial institutions handling private data.

Thailand is pouring nearly $2 billion into AI infrastructure and advanced electronics as global manufacturers shift supply chains into Southeast Asia. Japanese firm Datasection will build GPU server farms near Bangkok, while Taiwan Union Technology will manufacture specialized circuit board materials for AI data centers in Chonburi. These investments position Thailand as a regional hub for AI computing infrastructure.

Tech giants are pouring unprecedented sums into cloud infrastructure. Meta added $79 billion in new lease commitments (up 76%), while Microsoft added over $41 billion, pushing the combined total for major cloud players past $850 billion. Meta alone could spend $135 billion on capital expenditures in 2026. Much of this money funds AI data centers to support everything from Meta's social platforms to Microsoft's Azure cloud services. The spending spree is expected to continue over the next two decades.

Amazon is building AI tools to speed up film and TV production at its MGM Studios. The initiative, called AI Studio, will help with tasks like generating battle scenes (as shown in 'House of David' season two) and streamlining creative workflows. Human creators—writers, directors, actors—remain central, with AI handling repetitive or costly tasks. The move comes as production costs rise and follows recent layoffs partly attributed to AI efficiencies. A closed beta launches in March with industry partners.

Half of parents surveyed by Deloitte say they're worried their children rely too much on AI, while only 22% say their child's school provides approved AI tools. Nearly 30% of children are already using generative AI in schoolwork, yet a third of parents feel schools aren't teaching enough AI skills. Some parents are paying for private AI tutoring or camps to fill the gap.

OpenAI has launched GPT-Live, a new voice system for ChatGPT that can listen and speak at the same time — no more awkward pauses where you wait for the AI to finish. The system runs on "full-duplex" architecture, meaning it decides multiple times per second whether to talk, wait, or interrupt, just like a real conversation. For complex questions needing deep reasoning, it quietly hands the task to a smarter model while keeping the conversation flowing naturally. The feature is rolling out now to both paid and free ChatGPT users globally.

Nvidia has opened orders for its Vera processors to Chinese customers, potentially available by August. This marks a shift after US officials previously rejected H200 chip exports to China under a "presumption of denial" policy that changed to case-by-case review in January 2026. Chinese firms are reportedly planning to test hundreds of servers, with initial deployments at overseas data centers to access advanced AI chips.

Hong Kong-based Animoca Brands tested AI agents that can shop for users by accessing their Visa rewards, identifying relevant offers based on preferences, and completing purchases within set spending limits. The pilot with the Bruce Lee Club eShop in Hong Kong used Visa's payments infrastructure to enable agent-initiated transactions. The company says this could shift commerce from making human journeys faster to AI acting on users' behalf.

Meta's CTO Andrew Bosworth says the biggest AI mistake companies make is pulling future research into current product launches, leaving them unprepared for the next wave. Meta learned this the hard way with Llama: by loading Llama 3 with ideas meant for Llama 4, it fell behind rivals on reasoning and specialist AI capabilities. Bosworth argues that consumer AI won't take off unless it works quietly in the background without forcing people to change their habits — which is why Meta is betting on smart glasses to reduce phone reliance rather than add another screen.

Meta has paused its program that tracked employee keystrokes and mouse movements to train AI, after a researcher accidentally placed sensitive employee data somewhere it should not have been. The data — which was transformed — ended up accessible to more Meta staff than intended. Meta says there is no sign of deliberate wrongdoing. The program, launched in April, had already drawn employee backlash over privacy concerns and the lack of an opt-out option. Meta's CTO admitted employee morale is "probably one of the worst it's ever been."

Sprouts.ai, a California-based AI startup building 'Revenue Agents' for B2B sales teams, has raised $9 million in Pre-Series A funding. The round was co-led by True Global Ventures and Accel, with participation from the Philippines-based Kickstart Ventures. The startup's platform helps enterprises find, understand, and convert customers using AI workflows connected to systems like Salesforce. This signals continued investor appetite for AI tools that streamline sales and marketing operations.

The project will create more than 520 jobs, and source about 4.3 billion baht (US$130 million) of local agricultural inputs and raw materials each year.

Researchers at the University of California San Diego have achieved a world first: a humanoid robot successfully performed gallbladder removal surgery, both solo and in tandem with a second robot. Currently teleoperated by surgeons, the team aims for future autonomous capabilities. They say humanoids could bring critical surgeries to rural areas, battlefields, and space — at a fraction of the cost of current surgical robots.

Israeli optical chip startup DustPhotonics is in advanced talks to be acquired by Intel, Nvidia, or Amazon in a deal potentially worth hundreds of millions. The company's Carmel chips handle ultra-fast data transfer rates—critical infrastructure for AI data centers. Big tech is scrambling to lock up optical interconnect technology as copper wiring hits physical limits inside AI servers. The acquisition would let the buyer vertically integrate networking hardware with optical tech, cutting power use and data transfer delays.

France's competition authority is wrapping up its investigation into Nvidia, the dominant AI chipmaker. The probe examines whether Nvidia abused its market position in the AI computing space, following similar regulatory scrutiny in the US and EU.

A 21-year-old startup founder accidentally racked up $30,000 in AI token costs in a single month while building his AI learning app, mainly because he left a 'fast mode' setting on in Claude Code. He considers it worth the spend for speed, averaging $20,000 monthly on AI tooling. He notes his coding skills are 'atrophying' as AI takes over more of the actual writing work.

Article content unavailable for analysis.

Malaysia's tech sector is riding the AI wave, with analysts expecting continued growth through 2026 driven by data center and AI-related demand. Sustained spending from big cloud providers like Amazon, Google, and Microsoft is fueling semiconductor and tech infrastructure expansion across the country.

A Mendaki study found 87% of Malay/Muslim workers in Singapore are AI-literate. However, a worrying gap exists: technicians and associate professionals feel most at risk of AI replacing their jobs, yet have the lowest AI literacy and urgency to learn. Interestingly, older workers aged 46-60 show the highest motivation to learn AI and use AI tools most, while younger workers aged 18-30 show less urgency. National Development Minister Chee Hong Tat warned that AI benefits won't be evenly distributed and called on community groups like Mendaki to help ensure no one is left behind.

Personal data of around 70,000 people, including names, NRIC numbers, and past property addresses dating back to 1998, was exposed after a security lapse involving the Singapore Land Authority (SLA) and its vendor IBM. The data sat in development and testing environments without proper anonymisation. Cybersecurity firm Arctic Wolf warns this exposes a common blind spot: organisations often assume data security becomes the vendor's problem after signing a contract. In reality, the data owner bears the consequences of any breach. The incident highlights the need for continuous monitoring of where sensitive data lives and who can access it, rather than treating vendor security as a one-time checkbox.

Travel booking platform Skyscanner has rolled out AI tools letting users describe dream trips in plain language—like "cheap flights to Japan in December"—and get personalized recommendations comparing prices, weather and vibes. A new Road Trip Planner also builds itineraries matching car hires to routes like "Scenic" or "Adventure." Early tests show 60% of users clicked through to book after using the feature.

[IBM](https://www.ibm.com/) and [Red Hat](https://www.redhat.com/en) have launched Lightwell, a commercial platform designed to automate open source vulnerability remediation at scale, with cybersecurity firm [JFrog](https://jfrog.com/) joining as a launch technology collaborator. The platform arrives through two offerings: Lightwell Network, now generally available with a launch catalogue of more than 6,500 remediated, digitally signed and certified application-layer dependencies across Java, Python, JavaScript and other major ecosystems; and Lightwell Clearinghouse Premier, entering limited availability as a trusted intermediary for secured patch embargoes, initially restricted to the financial services sector. ## Built on a $5 billion open source security commitment The launch builds on

The UN's International Telecommunication Union has launched a new Focus Group to create global frameworks for trustworthy AI agents—systems that act independently on users' behalf, handling tasks from scheduling to financial transactions. The initiative responds to concerns about AI agents impersonating people and making unauthorized decisions. The group will develop standards to ensure AI agents remain identifiable and subject to human control, especially in sensitive areas like banking and critical infrastructure. Their first meeting is scheduled for November in Paris.

A new survey of 100 manufacturing leaders found zero reported significant revenue gains or cost savings from AI despite heavy investment. The problem isn't the technology—it's how companies buy it. Nearly half cite competitive anxiety as their main reason for spending, jumping straight to pilots without defining what number the project should move on the balance sheet. The result: 48% are stuck in pilot programs, versus 34% across other industries. Experts say manufacturers are buying AI tools then waiting for vendors to figure out deployment, rather than starting with a business problem and working backward.

Major news outlets including the New York Times and Daily News are asking a federal judge to sanction OpenAI for allegedly hiding evidence about how ChatGPT was trained on copyrighted journalism. The newspapers claim OpenAI has been obstructive for two years, "making misrepresentations" about its ability to search for copyrighted content in training data. The Times has already spent over $28 million fighting AI companies in court. This high-stakes copyright lawsuit could reshape how AI companies use news content and may cost OpenAI significantly in sanctions and legal fees.

Singapore-based AI tech services firm [NCS](https://www.ncs.co/en-sg/) has expanded its suite of enterprise-ready AI solutions, called Sunshine.AI, to include agentic assistants, developer tools, an AI foundational platform, productivity platforms, and an AI safety engine. With Sunshine.AI, they aim to help organisations accelerate transformation with safety, governance, and compliance from day one. Launched at its annual forum, NCS AI Impact 2026, the Singtel subsidiary also announced its partnership with [Alibaba Cloud](https://www.alibabacloud.com/). The partners will work to help organisations adopt AI securely while optimising performance, scalability, and cost efficiency. They aim to combine Alibaba Cloud’s AI infrastructure and intelligence, including Qwen’s AI platforms and foundat

A thought leadership piece exploring what AI might look like by 2036. The article envisions AI systems as interconnected ecosystems with contextual awareness and cooperative reasoning. It covers speculative topics like neuromorphic computing (brain-inspired chips), quantum AI, humanoid robots in healthcare and hazardous jobs, brain-computer interfaces, and AI-powered personalized education. The piece predicts AI will become civilization's "cognitive operating system" and raises questions about AI rights and consciousness. Much of this is forward-looking speculation rather than current technology.

SpaceX just went public at a $1.77 trillion valuation, and with Anthropic and OpenAI pushing into the trillions, these three companies together will generate more value than ALL US VC-backed exits since 2000. Last year, US IPOs raised just $70 billion in total — these three AI giants will dwarf that figure. It's the biggest wave of tech exits in history, driven by companies staying private longer and the capital-intensive nature of AI training.

Singapore tech firm NCS, a Singtel subsidiary, has launched an upgraded Sunshine.AI suite of enterprise tools including AI coding assistants, developer platforms, and safety engines. The company also partnered with Alibaba Cloud to help businesses adopt AI securely. NCS reports its tools boosted developer productivity by 15%, cut IT response times by 40%, and saved employees over 2 hours weekly. CEO Sam Liew said the firm is 'rewiring' from a systems integrator to an AI-led tech services organization.

The world's largest AI governance summit drew 12,000 people from 170 countries to Geneva to debate how AI should be regulated globally. This coincides with a first-of-its-kind UN dialogue involving 193 countries. The exhibition showcased everything from robot guide dogs for blind people ($4,000 each) to brain-computer interfaces helping stroke patients communicate, and wearable AI mobility supports. Experts warned we are entering a 'zero-click world' where AI agents act autonomously on our behalf without waiting for prompts.

A Brown University professor saw suspicious grade patterns when he switched from a take-home midterm to an in-person final exam. Students who scored in the 90s on the take-home test dropped to below 20% on the in-person exam. He suspects widespread AI cheating and says the incident is a 'wake-up call' for professors. The university is now investigating. This highlights growing concerns about how easily students can use AI to cheat on assignments.

Famous investor Michael Burry, known for predicting the 2008 housing crash, is warning that AI stocks face a fundamental contradiction. While companies like Microsoft, Meta, and Amazon need their massive AI spending to end in a few years, chipmaker Nvidia needs permanent endless demand for its expensive AI processors. Burry says there's no 'third door' that makes both outcomes possible, and he's betting against Nvidia and Tesla.

The RAISE Summit in Paris drew 9,000 attendees to discuss enterprise AI adoption, neocloud infrastructure, and energy challenges. French PM Macron pushed for European AI sovereignty, while investors like Mark Cuban highlighted how AI coding tools are accelerating startup creation. Key debates centered on open-source vs. proprietary models and data sovereignty concerns. Neocloud providers like CoreWeave and Vultr are evolving beyond GPU rental toward full-service infrastructure.

India's version of 'Who Wants to Be a Millionaire?' is integrating Google Gemini as an AI guide for viewers. The chatbot will help people register for the show and prepare for trivia questions. It's the latest example of tech companies partnering with entertainment shows to showcase AI capabilities to mass audiences through interactive features.

A Forbes contributor argues that as AI becomes standard in business, the real competitive edge won't be who has the smartest tech, but who uses AI most responsibly. The piece cites McKinsey data showing 78% of organisations have already embedded AI into at least one function. But consumers are wary — only 44% feel comfortable with businesses using AI in customer interactions, according to Edelman. The takeaway: AI doesn't create trust, it reveals whether your organisation already had it.

OpenAI's most powerful model yet, GPT-5.6, is now available to everyone after getting the Trump administration's approval. CEO Sam Altman calls it "the best model we have ever produced." The company also announced "ChatGPT Work" for business customers. Expect premium pricing for the new features.

OpenAI has launched ChatGPT Work, an AI agent designed to handle workplace tasks across multiple applications. Powered by the new GPT-5.6 model, it can work continuously on complex projects, gathering info from files and apps to produce finished documents, spreadsheets, and presentations. The company also unveiled a new desktop app. This marks a bigger push by OpenAI into automating office work as it prepares for a potential IPO.

OpenAI has launched ChatGPT Work, a powerful new agent feature that can independently handle complex workflows across apps like Google Drive, Slack, and Salesforce. Powered by the newly released GPT-5.6 model, it can break down big projects into steps, execute them in the background, and deliver finished materials like spreadsheets, presentations, and documents. Users can track progress, give feedback, and approve actions. It also includes scheduled tasks that run automatically even when you're away, and on desktop, it can control your computer directly to complete work.

When OpenAI releases a new powerful AI model, who actually checks if it's safe? According to experts, nobody really knows. The process is opaque — there's no clear agency, set of standards, or public oversight for deciding if advanced AI models can be released. Critics say this secretive approach, especially given reported donations to the Trump administration, raises serious questions about whether the public can trust how AI safety is being evaluated.

Energy companies are increasingly using AI-powered drones and robots to inspect pipelines, refineries, and other infrastructure instead of sending human workers into hazardous areas. At a recent industry summit, major oil and gas firms shared how these robots with special sensors monitor for corrosion, damage, and safety issues across vast geographical areas. The technology allows for constant surveillance and early problem detection while keeping workers safe.

Robotics startup 1X has unveiled major upgrades to its Neo humanoid robot's hands: 25 degrees of freedom, the ability to lift a 20-pound kettlebell yet pick up a single grape, touch sensors that detect slippage in real-time, and washable IP68-rated waterproofing. The hands use low gear ratios so the robot can 'feel' what it's touching, similar to humans. This makes Neo potentially the first robot that can actually do household chores like cooking and cleaning up after itself.

OpenAI just dropped GPT-5.6, its newest family of AI models, in three tiers: Sol (top dog), Terra (middle ground), and Luna (budget-friendly). The flagship Sol model is 54% more efficient at coding tasks than its predecessor and directly targets Anthropic's Claude Fable 5, claiming to beat it on speed, cost, and benchmark scores. OpenAI also launched ChatGPT Work, a workplace companion that can handle spreadsheets, presentations, and long projects across desktop, web, and mobile. Pricing ranges from $1 to $5 per million input tokens depending on the tier.