ChatGPT Wallet lets OpenAI's agents pay for things after you approve

✕Sat 14 Nov ‹›

AI agents · Sat 14 Nov

ChatGPT Wallet lets OpenAI's agents pay for things after you approve

1security check needed for every purchase

OpenAI launched ChatGPT Wallet on 13 November (US time), so its ChatGPT , called dots, can finish purchases for you. It starts with US ChatGPT Pro subscribers aged 18 and over who have an eligible US-issued Visa, Mastercard or American Express card.

Key points

How it works: you add a card once. Before each purchase you see the items, the total and the card, then approve with a check such as Face ID or a fingerprint. OpenAI says each approval leaves a record of what you allowed the agent to buy.
The agent never sees your real card number. After you approve, the card network issues a , a stand-in card number, so neither the dot nor the shop gets your actual details. OpenAI says card rewards and protections for eligible disputes still apply.
Shops may still block or decline agent purchases. OpenAI published a Commerce Partner Playbook for retailers and is working with payment firms such as Stripe, Adyen and Checkout.com and fraud firms such as Forter and Signifyd. TechCrunch notes Amazon bans outside AI agents from its store.
Later, OpenAI expects to let users set spending limits, product categories or time periods so the agent can buy within pre-approved rules. That is a plan, not a feature yet.

Why it matters: This is one of the first agent payment systems built with all three big US card networks, and for now a person still approves every purchase. It is not available in Malaysia yet; if it arrives, keep the per-purchase check and read each total before you approve.

Sources (2)

Introducing ChatGPT Wallet · OpenAI
ChatGPT's new wallet lets AI agents shop and check out for you · TechCrunch

Next · Singapore & MalaysiaSingapore bars ex-Megaspeed boss as Malaysia says it never cleared Megaspeed →

✕Sat 14 Nov ‹›

Singapore & Malaysia · Sat 14 Nov

Singapore bars ex-Megaspeed boss as Malaysia says it never cleared Megaspeed

917servers in Singapore's latest money-laundering charges

Singapore's Ministry of Home Affairs said on 12 November that Alice Huang, former chief executive of Singapore-based Nvidia partner Megaspeed, got a stern warning and is barred from re-entering Singapore. On 13 November Malaysia's trade ministry MITI said it had never cleared Megaspeed or its Malaysian affiliate Speedmatrix.

Key points

Singapore: after consulting the Attorney-General's Chambers, police found too little evidence that Huang committed an offence, so she was not charged. They say she knowingly gave inaccurate information to play down her role. She was not investigated for spying; police held her passport from June 2025 until 26 February 2026.
Why Malaysia is involved: the ministry says servers bought by companies under investigation went through Singapore to Malaysia without entering Singapore's market. Police are checking whether they were then sent on somewhere else, and say server maker Supermicro has given only limited help despite requests to US authorities.
Malaysia: MITI says its earlier visits were routine checks, not approval. With the Royal Malaysian Customs Department it is checking whether chips brought in for Speedmatrix were . Bringing AI chips in needs no permit, but exporting or re-exporting high-end US AI chips has needed a permit since July 2025.
In a separate Nvidia-chip case, Singapore charged a fifth person, Shi Bingzhe, on 13 November with 41 counts over 917 servers bought from Aperia International, worth more than US$100 million (about RM408 million), CNA reports. Megaspeed's chief executive says it is cooperating with US inspections.

Why it matters: Both governments are trying to show they are not for restricted US chips. Data centre operators and tenants in Malaysia should expect more questions about where every imported server ends up, and should keep permits and shipping records in order.

Sources (4)

Next · Cloud outageMicrosoft outage took Outlook, Teams and Copilot offline for 18 hours →

✕Sat 14 Nov ‹›

Cloud outage · Sat 14 Nov

Microsoft outage took Outlook, Teams and Copilot offline for 18 hours

18 hoursof downtime before traffic returned (Kentik)

A Microsoft 365 outage that began early on 13 November (Malaysia time) knocked out Outlook, Teams, Copilot and other services around the world. Microsoft says servers were failing and that it has found no evidence of malicious activity.

Key points

What broke: Microsoft said a core process on affected servers became overwhelmed, so servers failed health checks and dropped out; some failed again once traffic returned. It extended email timeouts to 48 hours so delayed mail would not be lost, and promised a full on the cause.
Who was hit: Malay Mail, citing AsiaOne, says OpenAI and Anthropic reported problems and Singapore's warned email replies could be delayed. Ars Technica, citing NBC, says Japan's government IT systems and the Philippines' Supreme Court were disrupted.
AI tools broke too: ITmedia reports ChatGPT's and Claude's Microsoft 365 stopped working, with Anthropic saying most mail, calendar and search requests failed. Ordinary chats outside those connectors were not part of the incident.
Traffic returned after about 18 hours of downtime, internet monitoring firm Kentik's Doug Madory said, according to TechCrunch. Microsoft's Copilot chief Charles Lamanna said teams are reviewing how to prevent a repeat.

Why it matters: AI assistants built into office tools are only as reliable as the cloud under them. Keep offline copies of key contacts and files, and agree a backup channel such as phone or WhatsApp for days when Outlook and Teams are down.

Sources (4)

Next · RobotsOpenAI breaks up robotics team, betting its main models can run robots →

✕Sat 14 Nov ‹›

Robots · Sat 14 Nov

OpenAI breaks up robotics team, betting its main models can run robots

74.8%Astra agent score on a simulated robot benchmark (Nvidia study)

OpenAI has disbanded its stand-alone robotics team and moved the staff into other groups, Fortune reported on 13 November (US time), and OpenAI confirmed the change. The company says robotics is still a core goal and the move is meant to speed it up.

Key points

Robotics researchers now sit in OpenAI's main research group under chief research officer Mark Chen. Applied robotics and hardware staff moved to the applied group under president Greg Brockman.
OpenAI's reason: its general-purpose models, including GPT-6 Astra, can now be applied directly to robots. Fortune cites an Nvidia study in which an Astra-powered scored 74.8% on a simulated robot , against 51.0% for the best specialised robot models. That is , not proof in real factories.
OpenAI is still building its own , with CEO Sam Altman targeting a first public demo in 2027. The change came so recently that OpenAI's booth at the CoRL robotics conference in Austin this week stood empty.
Not everyone agrees: Carnegie Mellon's Guanya Shi notes large models need a constant connection, so small models running on the robot still matter, and MIT's Russ Tedrake says robot makers' real-world data remains an advantage.

Why it matters: If big general models can control robots, factories and warehouses may not need a separate robot brain for every machine, which could cut costs. Treat the benchmark lead as early: it was measured in simulation, and robots on the floor still need control that works offline.

Sources (1)

Exclusive: OpenAI disbands its stand-alone robotics team as its general models start running robots · Fortune

Next · Developer toolsGoogle's Gemini Decision 1 answers yes-or-no questions without writing text →

✕Sat 14 Nov ‹›

Developer tools · Sat 14 Nov

Google's Gemini Decision 1 answers yes-or-no questions without writing text

1,000questions allowed in one request

Google launched the Gemini Decisions on 11 November (US time), powered by a new model, Gemini Decision 1. Instead of writing an answer word by word, it picks from options you give it and returns probabilities plus a from 0 to 1.

Key points

Three question types: yes/no, multiple choice (2 to 255 options) and a score on a scale you define. One request can ask up to 1,000 questions about a single input of text, images, audio, video or PDF.
Google's figures: US$0.10 per million input , nothing for output, and 150 to 250 milliseconds per request. Google says it is nearly 5x cheaper than Gemini 3.5 Flash-Lite on its suite; these are Google's own claims.
In a hands-on video on 13 November, the Prompt Engineering channel asked about the weather in Paris and the model picked 'technical support' with 92% confidence, because no option fitted. Adding an 'other' option fixed it: confidence only tells you which of your options is closest.
The model cannot explain its answers because it does not write text. The tester's rules: give every choice a way out, send only what each decision needs, ask small questions and test on your own data.

Why it matters: Many business tasks are small decisions, such as which team gets a , and a cheap, fast like this can handle them and pass unsure cases to a person. Set a confidence cut-off and keep a normal chatbot or a human for anything that needs a reason.

Sources (2)

Gemini Decisions API: Fast, calibrated multimodal decisions · Google
Gemini Decision 1 hands-on (Prompt Engineering) · Prompt Engineering

Next · Big TechJensen Huang quits as Nvidia CEO after affair and spy-link reports →

✕Fri 13 Nov ‹›

Big Tech · Fri 13 Nov

Jensen Huang quits as Nvidia CEO after affair and spy-link reports

33 yearsHuang led Nvidia from its founding in 1993 until 12 November

Nvidia said on 12 November (US time) that co-founder Jensen Huang has resigned as president, chief executive and board member with immediate effect, ending 33 years in charge. Chief financial officer Colette Kress becomes while the board searches for a permanent successor, according to Nvidia's statement as reported by Lowyat.net and TechNode Global.

Key points

The exit follows a week of reports. On 8 November Huang admitted an affair with Alice Huang (no relation), a former executive of Singapore-based firm Megaspeed whom US officials reportedly suspect of being a Chinese intelligence asset; Beijing denies she is a spy. On 11 November The New York Times reported a second affair, with Nvidia healthcare vice-president Kimberly Powell, who reported to him.
Nvidia also named former Synopsys chief executive Harvey C. Jones as its first-ever board chair. The board says operations, customer commitments and the product roadmap will continue without interruption, and it will consider internal and external candidates for CEO.
Nvidia shares fell more than 10% in , Reuters reported via Rappler. One fund manager told Reuters the product roadmap is visible for years; the risk is execution, because Huang personally coordinated deals with cloud firms, suppliers and AI labs.
Megaspeed, an official Nvidia cloud partner, says it is cooperating with US Commerce Department inspections and has hired a US law firm to count its . Singapore's Home Affairs Ministry says its own probe into servers exported via Singapore to Malaysia did not involve espionage, SoyaCincau reports.

Why it matters: Nvidia makes the chips most AI runs on, so who leads it affects supply, prices and the US export rules that reach Malaysia's boom. Watch who becomes permanent CEO and whether Washington tightens chip rules for South-east Asia.

Sources (7)

Jensen Huang Resigns As CEO And President Of NVIDIA · Lowyat.net
It's official: NVIDIA CEO Jensen Huang resigns · TechNode Global
Nvidia CEO Jensen Huang resigns amid affair and alleged Chinese spy scandal · SoyaCincau
Nvidia CEO Jensen Huang steps down in big blow to AI boom · Rappler (Reuters)
Megaspeed, Singapore-registered Nvidia cloud customer, cooperating with US export inspections · TechNode Global
Jensen Huang Is Out. Everything in AI Is Now Up for Grabs. · The AI Daily Brief
Jensen Huang just pulled out of Nvidia... · Fireship

Next · MalaysiaMalaysia drafts law to stop AI chips passing through to other countries →

✕Fri 13 Nov ‹›

Malaysia · Fri 13 Nov

Malaysia drafts law to stop AI chips passing through to other countries

4things data centre operators may have to disclose: identity, location, electricity use and GPU imports

Investment, Trade and Industry Minister Johari Abdul Ghani said on 12 November that Malaysia will check whether Speedmatrix Sdn Bhd, the Malaysian subsidiary of Singapore's Megaspeed, imported to China, Lowyat.net and TechNode Global report. He also said a new law is being drafted to track AI chips in .

Key points

Johari said Malaysia has had no formal request from foreign authorities about Speedmatrix, but would help any investigation. Buying GPUs for its own data centre would be no problem; re-exporting them in breach of restrictions would need a closer look.
The planned law, separate from the Digital Infrastructure (Data Centre) Bill, would make operators and disclose who they are, where their facilities are, how much electricity they use and how many GPUs they import, to stop of advanced chips.
The concern is firms that rented data centre space here, imported GPUs and shipped them on without using them. Johari said Malaysia has no specific law to stop such chip movements, so the ministry is checking whether existing laws can be used. Malaysia began controlling exports of US-origin AI chips last year.
Allegations about Megaspeed, summarised in Fireship's 12 November video from Culper Research and press reports, include renting Nvidia chips housed in Malaysia to Chinese firms. These are allegations, not findings; Megaspeed says it is cooperating with US inspections.

Why it matters: Malaysia wants data centre investment without becoming known as a for restricted chips, which could invite US penalties. Operators and their customers should expect more paperwork on where every GPU ends up.

Sources (3)

Malaysia To Review Alleged NVIDIA Chip Re-Exports, Draft Data Centre Law · Lowyat.net
Malaysia reviewing existing laws to prevent country from being GPU re-export hub - report · TechNode Global
Jensen Huang just pulled out of Nvidia... · Fireship

Next · RobotsGXO plans up to 1,000 Apptronik humanoid robots in its warehouses →

✕Fri 13 Nov ‹›

Robots · Fri 13 Nov

GXO plans up to 1,000 Apptronik humanoid robots in its warehouses

1,000Apollo robots over four years, only if agreed targets are met

Logistics company GXO said on 12 November (US time) that it has signed a with Apptronik to widen trials of the Apollo , aiming for up to 1,000 robots worldwide over four years if the robots meet agreed targets.

Key points

From 2027, Apollo will be tried in live warehouses on picking items, moving cases and totes, packing and sorting. Each job will be judged against productivity, safety and operational targets agreed by both companies.
The first deployment is planned for North America in late 2027. GXO would pay for the robots as a service, under a model, rather than buying them outright.
The 1,000 figure is conditional, not an order. The firms have not said which sites, how many robots start, or the price. GXO says Apollo finished a US pilot picking small consumer goods earlier this year.
GXO chief executive Patrick Kelleher told FreightWaves he expects humanoids to cost US$15 to US$20 an hour all-in within two years (about RM61 to RM81), HumanoidMegaHub reports. That is a forecast, not a measured result.

Why it matters: Humanoid robots are moving from demos to contracts with pass-or-fail tests. For warehouse operators, the hourly cost target is the number to watch.

Sources (4)

GXO and Apptronik expand partnership to scale humanoid deployment · GXO
GXO could deploy 1,000 Apptronik humanoid robots over four years · DC Velocity
Inside GXO's plan for up to 1,000 Apptronik humanoids · HumanoidMegaHub
GXO Plans to Deploy Up to 1,000 Apptronik Humanoid Robots · Machine Dawn

Next · SearchGoogle puts its Gemini 4 Hydrogen model into Search for everyone →

✕Fri 13 Nov ‹›

Search · Fri 13 Nov

Google puts its Gemini 4 Hydrogen model into Search for everyone

All usersGemini 4 Hydrogen is live in Google Search globally, Google says

Google said on 12 November (US time) that Gemini 4 Hydrogen is now available globally in Google Search for all users, calling it the biggest model upgrade to Search yet.

Key points

Google says Search can now handle long, multi-part questions, awkward typos and very specific requests, and follows instructions more precisely.
It leans on '': running many searches at once across the sub-topics of your question. Google says this links you to a wider range of websites than before.
Answers should be more visual, with more images, comparison tables and small custom tools, and Google says responses improve in every language AI Mode supports.
These are Google's own claims; the post gives no test results or figures.

Why it matters: Search is still how most people look things up, so this reaches far more people than a new chatbot. Check important AI answers against the linked sources, especially on health or money.

Sources (1)

Google Search gets a major upgrade with Gemini 4 Hydrogen · Google (The Keyword)

Next · AI agentsUnited Airlines will block AI agents from booking and managing flights →

✕Fri 13 Nov ‹›

AI agents · Fri 13 Nov

United Airlines will block AI agents from booking and managing flights

Check-instill needs a human to confirm bag contents, United says

United Airlines said on 11 November (US time) that it will stop from managing customers' reservations and buying flights, citing cybersecurity, privacy and legal risks, The Record reports.

Key points

United warns that sharing with a personal AI assistant can expose sensitive information and lead to unintended purchases, account changes or use of MileagePlus miles.
It says automated actions could also break US aviation rules, because check-in needs a human to confirm that bags do not contain banned items such as explosives.
The block covers United's own website and app and, separately, bookings made through aggregators and agencies. Bank of America told customers last week never to share user IDs and passwords with agents.
An identity-verification company boss quoted by The Record argues that blocking agents only delays the problem, and that verified identity are a better long-term answer.

Why it matters: If you use an AI assistant for travel, some airlines and banks may start refusing it. Never paste airline or bank passwords into an AI chat, and do sensitive steps like check-in yourself.

Sources (2)

United to block customers from using AI agents, citing cyber threats · The Record
ユナイテッド航空、外部のAIアシスタントによる航空券の予約などを禁止へ · ITmedia AI+

Next · AI businessAnthropic sets IPO price range that could value it at US$1.4 trillion →

✕Thu 12 Nov ‹›

AI business · Thu 12 Nov

Anthropic sets IPO price range that could value it at US$1.4 trillion

US$1.4tnvaluation at the top of the price range, fully diluted (TechCrunch)

On 10 November Anthropic, the company behind Claude, launched the for its stock market listing. It is offering 825 million shares at US$70 to US$78 each and has applied to list on Nasdaq under the symbol 'ANT'. At the top of the range it would be worth about US$1.4 trillion on a , TechCrunch and CNBC report.

Key points

The share sale could raise US$57.75 billion to US$64.35 billion (about RM234 billion to RM261 billion). Separately, AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC agreed to buy US$26 billion of non-voting shares at the price in a ; Google and Nvidia are putting in US$10 billion each.
CNBC's sources say pricing is planned for 18 November and trading for 19 November, with about 20% of shares for retail investors. The is below the US$2 trillion talked about earlier, but above the US$965 billion of Anthropic's May funding round.
The updated filing says reached US$81 billion by the end of October, about US$10 billion more than a month earlier; that is the company's own figure. The founders keep control through special shares that carry 50.1% of the votes.
In the roadshow video, transcribed by developer Simon Willison, co-founder Tom Brown says Claude is trained and served on four chip families: AWS Trainium, Google TPUs, Nvidia and AMD GPUs. Anthropic claims its Mythos model got Claude running on AMD chips in three days with almost no human help.

Why it matters: Nothing changes in Claude for users today. But this listing is a big test of how much investors still believe in AI, and the result could affect how easily AI companies raise money next year.

Sources (4)

Anthropic launches roadshow for proposed initial public offering · Anthropic
Anthropic sets IPO price range of $70 to $78 per share to reach an implied valuation of $1.404 trillion · TechCrunch
Anthropic sets price range of $70 to $78 in upcoming IPO · CNBC
Anthropic's IPO roadshow video · Simon Willison's Weblog

Next · AI toolsClaude can now join Google Meet calls and build shared spreadsheets →

✕Thu 12 Nov ‹›

AI tools · Thu 12 Nov

Claude can now join Google Meet calls and build shared spreadsheets

19 NovClaude Data switches on by default for Enterprise plans

At an event in San Francisco on 10 November, Anthropic added three new tools to Claude: Meetings, Whiteboards and Sheets, all in beta. It also made Claude Data, formerly Claude Dashboards, available to all paid plans.

Key points

Claude Meetings is for Team and and works only with Google Meet at launch. Claude joins as a visible participant, posts a notice in the chat, drafts the plan being discussed and answers questions in the meeting chat. It won't join calls with outside guests unless an admin allows it, and it leaves if one joins.
Claude Whiteboards, on all paid plans, turns a messy spoken plan into a shared board with a timeline, owners and dates. It can group retro notes by theme, then turn the board into a document or slide deck, or send it to Lucid, Miro or Zoho's Vani.
Claude Sheets, also on paid plans, builds a spreadsheet with tabs, formulas and drop-downs from a prompt such as 'make a hiring budget', and outlines each edit it makes. Anthropic says the features are simple for now; you can export to Excel or CSV.
For admins: on Team plans all three are on by default. On Enterprise plans Whiteboards and Sheets start off, and Claude Data turns on by default on 19 November. Settings are under Organization settings > Artifacts.

Why it matters: If your team already pays for Claude, try Sheets or Whiteboards on a small tracker before moving real work into it. Enterprise admins should check the Artifacts settings before 19 November.

Sources (4)

Introducing Claude Whiteboards, Claude Sheets, and Claude Data · Claude by Anthropic
Anthropic launches Claude Meetings to turn workplace discussions into plans and tasks · TechCrunch
Anthropic has its own AI-powered office suite now · The Verge
「Claude」に共同編集の「Sheets」「Whiteboards」 · ITmedia AI+

Next · RoboticsGoogle's Gemini Robotics 2.5 helps humanoids sort car parts in factories →

✕Thu 12 Nov ‹›

Robotics · Thu 12 Nov

Google's Gemini Robotics 2.5 helps humanoids sort car parts in factories

96%kitting success in the Mercedes-Benz trial (Google figure)

On 10 November Google DeepMind introduced Gemini Robotics 2.5, its most capable : an AI that sees, follows instructions and moves a robot's arms and hands. It is rolling out to partners such as Agile Robots, Apptronik and Boston Dynamics, not to the public.

Key points

In a trial on Mercedes-Benz production lines in Hungary and Germany, Apptronik's Apollo 2 picked metal brackets from bins and placed them into trays in the order each car needs. This job is called . Google says they succeeded 96% of the time.
DeepMind tested the model on more than 200 tasks, such as serving food with tongs and using a lab pipette, and an Apollo 2 tied a ribbon bow with its five-finger hands. Ars Technica reports success rates 30% to 100% higher than the previous version, and 20% to 30% faster, in lab tests.
Google also released ASIMOV-, an open test of tricks that try to talk a robot into unsafe actions. The data is on Hugging Face for other researchers to use.
Apptronik CEO Jeff Cardenas says the models improved dramatically after a few weeks of extra training and carried over to the Mercedes-Benz lines with very little drop-off.

Why it matters: Robots that can handle mixed parts on a real line are getting closer, which matters for Malaysia's electronics and car plants. This was still a pilot with Google's own numbers, so ask vendors for results on your own line before buying.

Sources (4)

Gemini Robotics 2.5 accelerates real-world general dexterity · Google DeepMind
DeepMind reveals Gemini Robotics 2.5 and open safety benchmark "ASIMOV-jailbreak" · Ars Technica
Google DeepMind releases new advanced robotics AI model and next-generation gene prediction model · SiliconANGLE
Google DeepMind rolls out Gemini Robotics 2.5 to partners · Tech in Asia

Next · Southeast AsiaSingapore funds 13 firms to protect AI agents from hackers →

✕Thu 12 Nov ‹›

Southeast Asia · Thu 12 Nov

Singapore funds 13 firms to protect AI agents from hackers

S$5.6mshared by 13 companies (about RM17 million)

On 10 November Singapore's Cyber Security Agency (CSA) named 13 companies that will share about S$5.6 million (about RM17 million) to build tools against AI-enabled and quantum computing risks, e27 reports. It also opened a new round, CyberCall 2026.

Key points

Several projects protect themselves. CloudsineAI will map how an attacker could hop from one agent to the next. Flexxon's X-PHY gateway will check traffic between agents and their tools for and ; its goal of over 95% detection is a design target, not a test result.
IdentityPlane is building one system to manage both staff and AI agent identities. StrongKeep, founded in 2025, is working with Singapore's research agency A*STAR on an AI '' for firms with fewer than 200 staff.
Two firms target future quantum risks: pQCee will build tools to test encryption against quantum-style attacks, and ST Engineering D'Crypt a quantum-ready encryptor for industrial control systems. CyberCall 2026 is open until 9 January 2027.
e27 notes the money is modest: split evenly it is about S$430,000 per company. The real test is whether infrastructure operators and small firms go on to buy the tools.

Why it matters: If your company is plugging AI agents into its systems, these grants show the risks regulators now worry about: attackers hopping between agents, hijacked tools and unmanaged agent accounts. Security startups in Malaysia can follow CyberCall 2026, but check the eligibility rules first.

Sources (1)

Singapore is paying startups to secure the AI agents it is racing to deploy · e27

Next · MalaysiaEPG tops out RM480 million data centre module factory in Johor →

✕Thu 12 Nov ‹›

Malaysia · Thu 12 Nov

EPG tops out RM480 million data centre module factory in Johor

800+jobs planned when fully running (EPG figure)

EPG said on 11 November it had finished the structural frame of its S$153 million (US$119 million, about RM480 million) factory in Senai Airport City, Johor. The factory will build ready-made units and is due to be completed in the first quarter of 2027.

Key points

The site covers 107,450 square metres, including 70,000 square metres of production space, in a in Iskandar Malaysia. EPG says it can employ more than 800 professional and technical staff once fully running.
EPG says production lines are designed to be at least 80% automated, with automated welding, a to track production and sensors that give real-time safety alerts. These are the company's plans, not yet in operation.
It plans to work with local suppliers of steel, electrical and mechanical systems and cables, and with Malaysian universities and training bodies on technical skills.
Modular data centres are built as factory-made units and put together on site. EPG is Singapore-based, with research centres in Singapore, Malaysia and Shanghai.

Why it matters: Johor's data centre boom, driven partly by AI demand, is bringing factory jobs as well as server halls. If you are an engineer or supplier in Johor, EPG says it will hire and buy locally once the plant opens in 2027.

Sources (1)

EPG Modular Data Center tops out $119M Malaysian factory at Senai Airport City · TechNode Global

Next · AI modelsGoogle opens Gemini 4 Carbon, its strongest model, to paying users →

✕Wed 11 Nov ‹›

AI models · Wed 11 Nov

Google opens Gemini 4 Carbon, its strongest model, to paying users

US$2 / US$10per million input / output tokens until 30 Nov, then US$4 / US$20 (Google)

On 10 November Google made Gemini 4 Carbon, which it calls its strongest model, available beyond the cybersecurity testers who got it two weeks earlier. Developers can use it through the paid Gemini , and Google AI Ultra subscribers can try it in the Gemini app, AI Mode in Search and , Google's AI coding tool. Most performance claims below are Google's own.

Key points

Until 30 November the API costs US$2 per million input and US$10 per million output tokens (about RM8 and RM41), the same as Gemini 4 Argon. After that it doubles to US$4 and US$20 (about RM16 and RM81). Google says Carbon uses more tokens than Argon because it explores and checks its work for longer.
Google says Argon is still the right choice for most coding and office work; Carbon is for very hard, multi-step jobs. It says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench at less than half the cost per task of the next best outside model. Google calls these independent tests; we have not seen the full tables.
Google is previewing '/swarm' in Antigravity, which turns one request into a team of you can watch at work. Carbon takes the planning and checking roles, and faster, cheaper models do the rest.
In a hands-on video, the creator of the Prompt Engineering channel, who says he recently joined Google and was given early access, left Carbon working for about 20 hours on a 3D model of a fort in Peshawar. A fresh AI reviewer scored it about 5 out of 10. His advice: ask for many rounds of work, a progress file and a separate reviewer.

Why it matters: If you're on Google AI Ultra or pay for the API, try Carbon on one long, hard task where checking matters, and compare its cost with Argon before switching. Most people won't need it for everyday writing or chat.

Sources (3)

Gemini 4 Carbon is now available · Google (The Keyword)
Gemini 4 Carbon is now available for developers · Google (The Keyword)
Gemini 4 Carbon: The Most Capable Model from Google · Prompt Engineering

Next · BusinessBusinesses can now claim and update their profile inside ChatGPT →

✕Wed 11 Nov ‹›

Business · Wed 11 Nov

Businesses can now claim and update their profile inside ChatGPT

1.2bnweekly ChatGPT users (OpenAI's own figure)

On 10 November OpenAI launched Businesses in ChatGPT, in beta. Through its Business Hub, a company can claim its listing, correct key details and add product feeds with prices and stock, so ChatGPT has current information when people ask about it. OpenAI says any eligible business located where ChatGPT is available can claim its listing.

Key points

Once verified, a business sees aggregated numbers: how often it is mentioned, how many people click through to its website, and common topics in conversations where its site is cited. OpenAI says it does not share people's chats with businesses, and only appear above a minimum number of conversations.
Business , chat assistants a business sets up to answer questions about its own products, are available only to eligible online shops and online-service businesses in the United States for now. Those chats are labelled and kept separate from ChatGPT's own answers.
Businesses can also pay to promote their Business Agent as the landing point for ChatGPT ads; sponsored placements are labelled. OpenAI says 1.2 billion people use ChatGPT each week, its own figure, and plans to expand to more countries later without a date.

Why it matters: If you run a shop or service in Malaysia, ask ChatGPT about your business to see how it describes you, then check in Business Hub whether you can claim the listing and fix prices, hours or products. The agent features are US-only for now.

Sources (1)

Introducing Businesses in ChatGPT · OpenAI

Next · AI modelsMeta's Muse Ember aims to make routine AI agent work cheaper →

✕Wed 11 Nov ‹›

AI models · Wed 11 Nov

Meta's Muse Ember aims to make routine AI agent work cheaper

US$0.10 / US$0.50per million input / output tokens, prompts up to 200k (Meta)

On 9 November (US time) Meta released Muse Ember, its smallest and cheapest model, built for the small steps repeat all day: calling tools, small coding jobs, sorting requests and reading long documents. On Meta's Model it costs US$0.10 per million input and US$0.50 per million output tokens (about RM0.40 and RM2).

Key points

That is the same headline price as OpenAI's GPT-6 Luna and Anthropic's Claude Haiku 5.5. Independent developer Simon Willison notes the price rises for very long prompts: above 200,000 tokens Ember costs US$0.30 in and US$1.50 out.
Meta says Ember matches Haiku's best score on the Artificial Analysis index at about a third of the cost per task, and beats every Luna setting. Meta's chart cites Artificial Analysis results dated 9 November, though its Ember costs are estimates, and several other test scores are Meta's own.
It is , reading text, images, audio and video, and takes up to 1 million tokens at once. It has six reasoning levels; in Willison's quick drawing test, higher levels did not reliably give better results, and the 'high' level failed to finish.
Meta says it tested Ember for chemical, biological, cyber and loss-of-control risks and found all below its own high-risk thresholds, and checked whether it follows instructions hidden in web pages. These are Meta's own assessments.

Why it matters: If you or your developers run AI agents on lots of small tasks, a cheaper model for the routine steps can cut bills. Test it on your own tasks first: cheap per token isn't always cheap per job.

Sources (3)

Introducing Our Smallest, Most Efficient Model: Muse Ember · Meta AI
Introducing Our Smallest, Most Efficient Model: Muse Ember (link blog) · Simon Willison
Meta releases Muse Ember (Japanese) · THE BRIDGE

Today

Friday 13 Nov

Thursday 12 Nov

Wednesday 11 Nov

ChatGPT Wallet lets OpenAI's agents pay for things after you approve

Key points

Singapore bars ex-Megaspeed boss as Malaysia says it never cleared Megaspeed

Key points

Microsoft outage took Outlook, Teams and Copilot offline for 18 hours

Key points

OpenAI breaks up robotics team, betting its main models can run robots

Key points

Google's Gemini Decision 1 answers yes-or-no questions without writing text

Key points

Jensen Huang quits as Nvidia CEO after affair and spy-link reports

Key points

Malaysia drafts law to stop AI chips passing through to other countries

Key points

GXO plans up to 1,000 Apptronik humanoid robots in its warehouses

Key points

Google puts its Gemini 4 Hydrogen model into Search for everyone

Key points

United Airlines will block AI agents from booking and managing flights

Key points

Anthropic sets IPO price range that could value it at US$1.4 trillion

Key points

Claude can now join Google Meet calls and build shared spreadsheets

Key points

Google's Gemini Robotics 2.5 helps humanoids sort car parts in factories

Key points

Singapore funds 13 firms to protect AI agents from hackers

Key points

EPG tops out RM480 million data centre module factory in Johor

Key points

Google opens Gemini 4 Carbon, its strongest model, to paying users

Key points

Businesses can now claim and update their profile inside ChatGPT

Key points

Meta's Muse Ember aims to make routine AI agent work cheaper

Key points