Hackers used fake job tests and AI tools to rob Binance

✕Wed 4 Nov ‹›

AI and security · Wed 4 Nov

Hackers used fake job tests and AI tools to rob Binance

RM25.6 billionstolen from Binance on 29 Oct (about US$6.3 billion)

Binance says the record theft of about RM25.6 billion (US$6.3 billion) on 29 October began months earlier with a fake AI coding tool and a fake job interview. In the same week, a tampered Claude Code sat in Cursor's app store for about 11 hours.

Key points

How it started: in June, a Binance engineer looked for a popular AI coding tool and installed a tampered copy from a fake download page. In September, in a trick, fake recruiters gave another engineer a coding test with hidden . The attackers then moved through Binance's own internal AI tools to the system that approves money transfers (Binance's report, via CNBC).
Binance says it has put back 98% of the money from its own funds, and about RM11.6 billion (US$2.85 billion) has been frozen or seized. The FBI blames a North Korean hacking group called TraderTraitor. North Korea denies it.
On 29 October, a tampered copy of the Claude Code extension was live in Cursor's extension store for about 11 hours and reached at most 5,000 users. It tried to plant a hidden into their computers, a classic . Removing it is not enough: Cursor says to update to version 2.1.314 or later (Cursor, WIRED).
AI also finds faster. Anthropic says an unreleased Claude model found and used a Linux bug that gives an attacker , or full control, on some servers, in about 53 minutes by Fireship's count. A fix is out.
Close to home: Prime Minister Anwar warned on 3 November that Malaysia could be a target for cyberattacks, and Bank Negara is meeting the banks about it (Lowyat.net, citing The Edge).

Why it matters: Attackers now go after the tools people install to use AI. Download AI tools only from the official site, be wary of coding tests from recruiters you don't know, and keep updates switched on.

Sources (6)

Binance co-CEO Teng says hackers used fake job interviews to steal $6.3 billion in crypto · CNBC
North Korea responsible for $6 billion Binance attack · FBI
Claude Code extension security notice · Cursor
Trojan malware targets thousands of Anthropic Claude developer systems · WIRED
Fireship on AI-found Linux bugs and recent hacks · Fireship
Anwar: Malaysia could be a target for cyberattacks · Lowyat.net

Next · How companies use AIA landlord, a security firm and Anthropic measure AI's payoff →

✕Wed 4 Nov ‹›

How companies use AI · Wed 4 Nov

A landlord, a security firm and Anthropic measure AI's payoff

99%lower cost per listing photo at Heimstaden in the Netherlands

A European housing landlord cut the cost of listing photos by 99%. Norton's maker built a news app three times faster, and Anthropic rewrote a core system in a language its team didn't know.

Key points

Heimstaden (rental housing, Europe): a listing photo with furniture added by computer () used to cost about RM228 (€50). With AI it now costs under RM0.46 (€0.10), which the company estimates could save about RM590,000 (€130,000) a year in the Netherlands alone (Heimstaden, in OpenAI's case study).
At Heimstaden, an analyst with no coding background used to check 20,112 properties for climate risk. The company says this took 2 people two months, against an estimated 5 people for a year. Half its staff used ChatGPT within 2 to 3 months of launch.
Gen Digital (maker of Norton): a small team built the Norton Revamp news app, launched in March 2026, with AI coding help. It shipped features 3 times faster and about 70% of the screen code was AI-written. But checking the AI's work took 20 to 30% of their time (Gen's engineers, AWS blog).
Anthropic: a team with no experience used Claude to move a core part of its from Python to Rust. fell 40 to 50% and first replies came 10 to 20% faster, says Anthropic's Katelyn Lesse (InfoQ).
Lesson for small firms: the biggest savings come from repeated jobs, like photos or property checks. Budget real time for someone to check what the AI produces.

Why it matters: These are measured results, and Norton's team also counted the cost of checking AI's work. That hidden cost is what a small business needs to plan for.

Sources (3)

Heimstaden scales AI across its business with ChatGPT and Codex · OpenAI
How Gen Digital built Norton Revamp's AI-curated news engine with Amazon Bedrock AgentCore · AWS Machine Learning Blog
Anthropic moves its Messages API from Python to Rust with Claude (Chinese) · InfoQ China

Next · Using AI wellHow to stop sloppy AI work from wasting your colleagues' time →

✕Wed 4 Nov ‹›

Using AI well · Wed 4 Nov

How to stop sloppy AI work from wasting your colleagues' time

~2 hourslost per piece of sloppy AI work received (Stanford/BetterUp)

Sloppy AI-made work, called '', costs the people who receive it hours to fix. The fix is simple: give the AI your own view and some before it writes anything.

Key points

A Stanford and BetterUp study found about 40% of US office workers had received workslop in the past month, and each case took close to two hours to deal with (The AI Daily Brief).
Four checks before you send AI-made work: could you defend every sentence? Did you read all of it? Does it include something only you could add? Would you put your name on it?
Say what you think first. The same plain prompt gave almost identical LinkedIn posts in ChatGPT, Claude and Gemini. Adding the writer's own opinion, and words to avoid, gave a much better draft.
Give it context once: Notion shows a setup that writes an 'operating guide' page. You paste it into ChatGPT's or Claude's , so the AI knows where your projects and notes are, and you can save repeated jobs as (Notion).

Why it matters: AI makes drafts cheap, but the reader pays for the checking. A few minutes adding your view saves everyone else time and protects your name.

Sources (2)

How to Deslopify Your AI · The AI Daily Brief
Set up ChatGPT and Claude with your Notion workspace · Notion

Next · AI and creatorsEU court adviser: AI answers copying news articles can break copyright →

✕Wed 4 Nov ‹›

AI and creators · Wed 4 Nov

EU court adviser: AI answers copying news articles can break copyright

6,000+new game pages a day on Itch.io, double the August level

An adviser to the EU's top court says Google's Gemini may break publishers' rights when its answers stand in for the original article. Separately, the game store Itch.io deleted terabytes of AI-made junk.

Key points

In the EU court's first case on generative AI, Hungarian publisher Like Company v Google, Szpunar said on 3 November that Gemini may break the rules if it gets past or bot blocks, or if its answer copies an article so closely that readers no longer need the original (CJEU press release).
Facts are free to use; only the way a story is written is protected. He also said training AI on articles is generally allowed under the EU's . His opinion is not binding, and the judges rule later (Courthouse News, RTÉ).
Itch.io: new game pages a day doubled from about 3,000 to more than 6,000 between 3 August and 30 October, but downloads did not rise. The store deleted 'a couple terabytes' of and added a 'No AI' search filter (Kotaku).
The filter only shows games whose makers say they used no AI, so most games now drop out of it. Makers of tabletop games, books, comics and soundtracks must now say whether they used AI.

Why it matters: Courts and stores are starting to set limits on AI reusing people's work. If you use AI to write or make things, credit and link the original, and say when AI helped.

Sources (4)

Advocate General's Opinion in Case C-250/25 Like Company v Google (press release) · Court of Justice of the EU
EU legal adviser finds limited uses of Google AI could run afoul of publishers' rights · Courthouse News
EU court adviser backs publishers in Google AI case · RTÉ
Itch.io deletes terabytes of gen-AI slop and adds a No AI filter · Kotaku

Next · Southeast AsiaPenang firm opens RM20 million plant for AI chip inspection machines →

✕Wed 4 Nov ‹›

Southeast Asia · Wed 4 Nov

Penang firm opens RM20 million plant for AI chip inspection machines

RM20 millionEstek's new AI and chip-equipment plant in Batu Kawan, Penang

A homegrown Penang company opened a new plant that builds chip-making machines and AI inspection tools. Singapore's NCS is tripling its AI deployment engineers, and a Thai bank's fund backed AI group Amity.

Key points

Malaysia: Estek Automation, a Penang company started in 2015, opened a 50,000 sq ft plant in Batu Kawan worth RM20 million. It makes chip-making machines such as die sorters and , and AI camera checks that spot faulty parts (The Star, Digital News Asia).
Singapore: NCS opened an AI Mission Centre to help organisations move from AI trials to daily use. It plans to grow its embedded AI engineers from 100 to more than 300 in two years, and signed deals with Changi Airport Group, the Institute of Mental Health and Life3 Biotech (Digital News Asia, CNA).
Thailand: Krungsri Finnovate, the venture arm of Krungsri bank, invested in Amity, extending its Series D round past about RM407 million (US$100 million). Amity serves more than 100 business partners and thousands of small businesses, and is eyeing a stock market listing (TechNode Global, e27).

Why it matters: In the region, AI money is going into factories and the people who put AI to work, not only chatbots. That means jobs for engineers and technicians, not just coders.

Sources (6)

Estek invests RM20mil in new Batu Kawan manufacturing facility · The Star
Estek Automation opens US$5 mil manufacturing facility in Penang · Digital News Asia
Singapore's NCS launches AI Mission Centre, commits to over 300 forward-deployed engineers · Digital News Asia
NCS opens AI Mission Centre in Ang Mo Kio, to triple embedded AI engineers to over 300 · CNA
Krungsri Finnovate invests in Thailand's AI company Amity, Series D funding exceeds $100M · TechNode Global
Krungsri buys into Amity as the Thai AI group dresses its cap table for an IPO · e27

Next · Spotting AI fakesGoogle opens its AI-made photo and video checker to everyone →

✕Tue 3 Nov ‹›

Spotting AI fakes · Tue 3 Nov

Google opens its AI-made photo and video checker to everyone

180 billion+images and videos already watermarked by SynthID

Google has opened Detector to people around the world, in English. You upload an image, video or audio clip, and it tells you if it carries a hidden AI label from Google or its partners.

Key points

How it works: Google and partners hide an invisible watermark inside AI-made files. Google says it has marked more than 180 billion images and videos and 240,000 years of audio this way (Google's October recap).
It now also checks files made with tools from OpenAI, NVIDIA and Kakao, with Apple support expected soon. Before, the Gemini app could only check content made by Google's own AI (InfoQ).
It is not a general . It cannot spot files from Meta or Microsoft, which use different watermarks, and heavy editing can weaken the mark. A 'no clear signal' answer does not prove a file is real.
Critics on Reddit say a light clean-up of an image can strip the mark. Another approach, , saves a file's history as , but that is lost in screenshots and many uploads.
Also from Google: Android's AI scam detection already flags more than 5 million suspected scam calls and texts a month. A new opt-in Enhanced protection mode in Google Messages uses and Gemini to catch scams that unfold over a chat.

Why it matters: Fake photos, voices and videos are a common scam tool. A quick check helps, but a clean result is not proof, so keep checking where a file came from.

Sources (2)

The latest AI news we announced in October 2026 · Google
SynthID Detector launches worldwide, checks content from OpenAI, NVIDIA and Kakao (Chinese) · InfoQ China

Next · How companies use AICircle K and Mercari share measured results from using AI →

✕Tue 3 Nov ‹›

How companies use AI · Tue 3 Nov

Circle K and Mercari share measured results from using AI

Under 15 minCircle K's average incident triage time, down from 45-60 min

A global convenience store chain cut the time its engineers spend working out what broke. A Japanese marketplace got every employee using AI, then learned that tools alone were not enough.

Key points

Circle K (convenience and fuel stores, 17,200+ stores in 27 countries): when an alert fired, an spent 45 to 60 minutes pulling logs and checking systems before any fix could start. An built on now does that checking and posts a short summary in Slack within minutes.
Result: average time fell to under 15 minutes, a 67 to 75% cut, based on the team's own incident-response numbers (AWS blog, co-written with Circle K engineer Marek Klatowicz).
Mercari (online marketplace, Japan, about 2,000 staff): by the end of January, 100% of employees used AI tools like ChatGPT and Claude. Its earnings deck says output per engineer is 7.6 times what it was in the year to June 2024 (ITmedia).
But faster coding alone did not lift business results, says Toshiya Kimura. Mercari now builds AI agents for each stage of work, from planning to checks. Some managers blocked staff from spending time on agents, so Kimura told everyone every two weeks that this time was not wasted.
Lesson: AI pays off when it takes over a whole repeated job, like Circle K's checks, and when bosses make time for people to build it.

Why it matters: Both firms measured real time saved, not just how many people log in. Handing AI one boring, repeated task end to end is often where the gain shows up.

Sources (2)

How Circle K accelerated DevOps incident triage to under 15 minutes with Amazon Bedrock · AWS Machine Learning Blog
Who builds an AI-first organisation? Mercari's practice (Japanese) · ITmedia Business

Next · AI subscriptionsChatGPT and Codex will share one usage allowance from December →

✕Tue 3 Nov ‹›

AI subscriptions · Tue 3 Nov

ChatGPT and Codex will share one usage allowance from December

Decemberwhen ChatGPT and Codex start sharing one allowance

OpenAI is folding ChatGPT Work's agent features into the main ChatGPT app. In December, ChatGPT and , its coding helper, will draw from one shared usage allowance.

Key points

OpenAI announced this on 1 November. It says most people should not have to wait for their allowance to reset more often than now. Quick 'Instant' answers use none of it (ITmedia).
How much you use depends on the model and the task. You can see what is left, and when it resets, under Settings, then Usage.
On 29 October, OpenAI also doubled usage on its Pro 100, Pro 200, Pro 500 and Business Premium plans, and dropped a planned cut to Pro 200. Some Pro 200 users said calling that a 'doubling' was misleading, and Plus users got no increase (InfoQ).
Hidden background work can eat allowances too. One developer measured Claude Code's status summaries at 16.4% of a week's estimated cost. Anthropic says the average is about 2%, and fixed the costliest case in version 2.1.311.

Why it matters: If you pay for ChatGPT, chat and coding will soon come out of the same pot. Check the Usage page before a big job so you are not cut off halfway.

Sources (2)

ChatGPT to merge Chat and Work; usage limits shared with Codex (Japanese) · ITmedia AI+
OpenAI doubles Pro limits; background agents quietly use tokens (Chinese) · InfoQ China

Next · AI moneyAnthropic's sales are 12 times last year's, but it still loses money →

✕Tue 3 Nov ‹›

AI money · Tue 3 Nov

Anthropic's sales are 12 times last year's, but it still loses money

RM69 billionAnthropic's sales in July to September (about US$17 billion)

Anthropic, the maker of Claude, shared its July to September numbers ahead of its stock market listing. Sales soared, losses stayed huge, and China's Kling AI video unit closed a giant funding round.

Key points

Anthropic made about US$17 billion (about RM69 billion) in sales in the quarter, up about 1,127% from a year earlier. Growth is slowing, though: 47% over the previous quarter, down from 144% the quarter before (CNBC).
Its was US$23.3 billion (about RM95 billion). Most of that is an accounting charge tied to its rising value, not cash spent. Its , from actually running the business, was US$567 million (about RM2.3 billion) (Reuters, CNBC).
It has promised to spend about US$895 billion (about RM3.6 trillion) on computing and . Some investors value the company at about US$1.5 trillion (about RM6.1 trillion), Reuters reports.
In China, Kuaishou's Kling AI video unit completed RMB19.8 billion (about RM12.1 billion) in funding. Investors can ask for their money back, plus 8% a year, if Kling has not listed by October 2031 (TechNode Global).

Why it matters: AI tools are selling fast, but building them still costs more than they earn. That is why prices, plans and keep changing for users.

Sources (4)

Anthropic reports $17 billion in Q3 revenue, purchase commitments of $895.4 billion in revised S-1 · CNBC
Anthropic's loss shrinks, revenue jumps ahead of blockbuster IPO (Reuters) · Investing.com
Anthropic files Q3 numbers ahead of Nasdaq listing (Chinese) · IThome
Kuaishou's Kling AI completes $2.9B financing · TechNode Global

Next · RobotsRobots race up China's Mount Tai carrying eggs and watermelons →

✕Tue 3 Nov ‹›

Robots · Tue 3 Nov

Robots race up China's Mount Tai carrying eggs and watermelons

43 min 41 swinning robot dog time up Mount Tai; most tourists take 2+ hours

About 90 teams sent robots up more than 3,000 stone steps on one of China's most famous mountains. did well. Some fell, overheated or had to be carried down.

Key points

The course was 3.9 km long and climbed more than 600 metres. Robots either raced to the top or carried fragile goods like fresh eggs and watermelons (AFP, via Malay Mail).
A robot dog from Unitree's Beijing Lingyi team won the four-legged race in 43 minutes 41 seconds, including a 2-minute penalty, running on its own. Most tourists take at least two hours (MyDrivers, CCTV).
The fastest humanoid finished a shorter course in 27 minutes 26 seconds. One humanoid overheated halfway and was carried off on a stretcher, and another was led up on a rope.
Also: AGIBOT says 110 of its humanoid robots performed in sync and set a Guinness World Record. Unitree's July to September sales rose 70% to 665 million yuan (about RM405 million), but profit fell 17% as research spending jumped.

Why it matters: Stairs, slopes and heat are where robots still struggle. Races like this show what works outside the lab, and robot dogs are ahead of humanoids for now.

Sources (4)

Robots race up China's sacred Mount Tai carrying eggs and watermelons · Malay Mail
Unitree robot dogs win 1 gold and 2 silvers on Mount Tai (Chinese) · MyDrivers
110 humanoid robots. One synchronized performance. A GUINNESS WORLD RECORDS. · AGIBOT
Unitree Q3 revenue 665 million yuan, up 70% (Chinese) · MyDrivers

Next · Southeast AsiaGoogle pays RM650 million for Negeri Sembilan water near its data centre →

✕Tue 3 Nov ‹›

Southeast Asia · Tue 3 Nov

Google pays RM650 million for Negeri Sembilan water near its data centre

65 million litresof treated water a day from the new Negeri Sembilan plant

Google will fund a reservoir and a water treatment plant in Negeri Sembilan. They will serve its planned in Port Dickson and also add to the state's public water supply.

Key points

Google is putting in US$160 million (about RM650 million). Gamuda will build the plant, which can treat 65 million litres a day. The state body MBINS will own both facilities (Lowyat.net, The Sun).
The reservoir sits by the Linggi River. It stores extra river water and rain in wet seasons, which also helps reduce flooding, and supplies the plant in dry spells.
Gamuda said in January 2025 that 25% of the plant's capacity would go to homes and 75% to data centre use (The Sun). Google says it will also pay for all the electricity its campus uses and its grid connection.
Regionally, the Quad (Australia, India, Japan and the US) plans more than US$85 million (about RM345 million) in help to spread trusted AI across the Indo-Pacific. It has not said who gets it or when (TechNode Global).

Why it matters: AI data centres need huge amounts of water and power. Who pays for it, and whether locals benefit too, matters for every Malaysian state hosting one.

Sources (3)

Google invests US$160 million in Negeri Sembilan water infrastructure · Lowyat.net
Google to invest US$160 million in Negeri Sembilan water infrastructure · The Sun Malaysia
Quad plans $85M for trusted AI in the Indo-Pacific · TechNode Global

Next · AI and the webOpenAI, Google and Microsoft back one 'don't train on my site' switch →

✕Mon 2 Nov ‹›

AI and the web · Mon 2 Nov

OpenAI, Google and Microsoft back one 'don't train on my site' switch

1 linein robots.txt to say no to AI training, once supported

Websites may soon get one simple way to tell every big AI company not to use their pages to train AI. OpenAI, Google and Microsoft said on 30 October that they will support a shared signal, and Anthropic is already building it in.

Key points

How it works: a site owner adds a short rule to , the file that already tells where they can go. A new line, User-Agent: __usage_ai-training, followed by Disallow means 'don't train on this', and Allow means training is fine. You can also mix them, for example blocking the whole site but allowing a press kit folder (ITmedia, IThome).
It only covers training. It does not stop crawlers from visiting or search engines from showing your pages. If a site adds nothing, that counts as no stated preference, not a yes or a no.
Timing differs. Anthropic is first. OpenAI aims for the next few months, Google and the others aim for 2027, and Microsoft says it will start reading the signal on 1 February 2027. Google will also treat a 'don't train' signal as a no for its own Google-Extended setting unless a site says otherwise. Meta said it supports the idea but gave no plan.
It is still a proposal. The idea comes from an written by Anthropic's Kevin Kelley, which has no official standards status yet. Google says the switch also fits the EU AI Act's code of practice, which treats robots.txt as a machine-readable way to say no.

Why it matters: Today every AI company has its own , so site owners have to chase each one. One shared switch would make saying no to AI training a single line, if the companies follow through.

Sources (2)

Next · How companies use AILi Ning runs 50 AI agents as Google Labs rethinks planning →

✕Mon 2 Nov ‹›

How companies use AI · Mon 2 Nov

Li Ning runs 50 AI agents as Google Labs rethinks planning

460,000hours of equivalent work a year from Li Ning's AI agents

A Chinese sportswear brand shared how much work its now do, and a Google Labs product leader explained how her team decides what to build now that building is the easy part.

Key points

Li Ning (sportswear, China): about 50 AI agents now work in its supply chain and merchandise teams on WorkBuddy, Tencent's office agent. Together they do about 460,000 hours of equivalent work a year, two-thirds of it from one 'digital store manager' agent. Li Ning plans 200 agents within a year (China Times, via Pandaily).
Behind it, Tencent Cloud says AI agents now create more than 20,000 of its databases a day, about 80 times more than a year ago, with over a million in use. Features like and let an agent test a change on a copy before it goes live. WorkBuddy has handled 9 billion tasks since March.
Google Labs (US): Jaclyn Konzelmann says the question used to be 'can we build it?' and is now 'should it exist?'. Instead of a 40-page , her team starts with three things: a short strategy doc on what and why, a presentation to test the idea on colleagues, and a to see if people get excited (Peter Yang's interview).
Lesson: when agents can copy a simple product over a weekend, the code is not what lasts. Konzelmann's team looks for new model abilities and new habits people are forming, then builds products, not features.

Why it matters: Agents are moving from trials to real headcount-sized work, and planning is changing with them. A short strategy, a demo and a prototype now beat a long spec for testing an idea fast.

Sources (2)

AI agents now create over 20,000 Tencent Cloud TDSQL-C databases a day · Pandaily
How Google Labs decides what to build (with Jaclyn Konzelmann) · Peter Yang

Next · Southeast AsiaMalaysia's JurisTech brings its AI lending software to Thailand →

✕Mon 2 Nov ‹›

Southeast Asia · Mon 2 Nov

Malaysia's JurisTech brings its AI lending software to Thailand

28 OctJurisTech and TBN announce their Thailand partnership

Kuala Lumpur-based JurisTech has teamed up with TBN, a company listed in Thailand, to sell AI-powered lending software to Thailand's , the firms that lend to people and small businesses outside the big banks.

Key points

The deal, announced on 28 October, covers the whole loan journey: signing up customers, checking whether they can repay, managing loans, and collecting and recovering money (Digital News Asia).
AI does the slow paperwork, like reading documents and scoring risk, so lenders can decide faster. The companies stress human oversight: people still make the lending decisions.
Non-bank lenders often serve borrowers with little credit history, which is where faster, data-driven checks can help most. JurisTech's CEO See Wai Hun and TBN's CEO Panayu Sirikrachangsri both framed it as the start of a longer partnership.
It is an early-stage deal, so there are no results to show yet. It is one more example of a Malaysian AI software firm growing by selling into a neighbouring market.

Why it matters: Malaysian AI companies are starting to export practical tools, not just use them at home. Lending is a good test because better checks can mean fairer access to loans for small businesses.

Sources (1)

Malaysia's JurisTech partners Thai-listed TBN to target Thailand's non-bank lending market · Digital News Asia

Next · AI safetyA ChatGPT bug let one shared link steal account data →

✕Mon 2 Nov ‹›

AI safety · Mon 2 Nov

A ChatGPT bug let one shared link steal account data

RM20,300bug bounty OpenAI paid (US$5,000)

Security researchers showed how a single shared ChatGPT link could have taken over a signed-in user's account. OpenAI has fixed it. Separately, Anthropic reportedly told Claude not to copy other people's games.

Key points

Parameter Labs found a chain of small gaps in ChatGPT's HTML previews. A math-input tool inside the preview could be tricked into running outside code, so opening a booby-trapped shared link could leak a signed-in user's private data (). Nothing needs doing now, because the gap is closed (Parameter Labs, via Simon Willison).
OpenAI paid a US$5,000 , about RM20,300, which Simon Willison called low for a full account takeover.
Claude Code: according to QbitAI, Anthropic staff confirmed two sentences were added to the Opus 5.5 so Claude steers away from recreating someone else's specific game. It follows a wave of AI-made game clones and since Opus 5.5 came out in September.
Lesson: AI tools are now powerful enough that the risks are less about the chatbot saying something wrong and more about what it can be tricked into doing, or helped to copy.

Why it matters: As AI apps run more code for you, one bad link can matter more. Companies are fixing these gaps with bounties and also with quiet rules inside their models.

Sources (3)

How we took over ChatGPT accounts · Simon Willison
We found a gap at OpenAI · Parameter Labs
Claude Code told not to recreate other people's games (Chinese) · QbitAI

Next · AI in ChinaChina now uses AI more than it trains it, report says →

✕Mon 2 Nov ‹›

AI in China · Mon 2 Nov

China now uses AI more than it trains it, report says

175 trillionAI tokens used per day in China (June 2026)

A Chinese government research centre says AI use in China has passed a turning point: more computing now goes to answering people than to training new models. Alibaba's AI glasses and a top AI founder added to the picture.

Key points

CCID's report says , the work an AI does when it answers you, has passed training for the first time. China used more than 175 trillion a day as of June 2026, over 1,000 times early 2024 (IThome).
The same report counts more than 500 advanced smart factories and over 700 million generative AI users, more than half the population.
Alibaba's Qwen AI glasses got a free update that translates phone calls and videos as you watch or listen (IThome).
Tang Jie, founder of Z.ai (Zhipu), told the SCMP that , AI that helps improve itself, is a key step toward human-level AI, and that doing it safely is the main challenge.

Why it matters: When most AI computing goes to daily use rather than training, the race shifts to cheap, fast everyday answers. China's huge user base is making that shift first.

Sources (3)

CCID: China's AI inference data overtakes training data (Chinese) · IThome
Qwen AI glasses update adds call and media translation (Chinese) · IThome
Self-improving AI a key step towards AGI, safety is key: Z.ai founder · South China Morning Post

Next · AI toolsClaude's top plan drops its five-hour limit for verified users →

✕Sun 1 Nov ‹›

AI tools · Sun 1 Nov

Claude's top plan drops its five-hour limit for verified users

10xUltra's weekly usage compared with the Pro plan

Anthropic renamed its two biggest Claude plans and removed the five-hour on the top one, Ultra. That limit used to stop heavy users in the middle of long jobs, even when they still had usage left for the week.

Key points

New names, same prices: Max 5x is now just Max at about RM406 (US$100) a month, and Max 20x is now Ultra at about RM811 (US$200) a month. Subscribers don't need to do anything (Claude Help Center, The New Stack).
On Ultra, verified accounts no longer have the rolling five-hour session limit. Anthropic checks accounts automatically using account and payment history. Newer accounts keep a temporary limit and get an email when it is lifted. You can't ask for it to be sped up.
The weekly limit has not changed. Ultra gives 10 times the Pro plan's usage each week, and Max gives 5 times. Chat and Claude Code share the same pool, and both plans now include monthly credits equal to their price.
Anthropic's Mike Overell apologised that the old '20x' name confused people, because it never meant 20 times the weekly usage. OpenAI had doubled its own Pro plan's usage the day before (QbitAI).

Why it matters: If you use Claude for long tasks like coding or reports, the annoying stop halfway through goes away, but the weekly cap still decides how much you really get. AI companies are now competing on , not just on price.

Sources (3)

Anthropic just removed one of Claude's biggest usage limits, but not for everyone · The New Stack
What is the Ultra plan? · Claude Help Center
Claude Max renamed; Ultra removes the 5-hour limit (Chinese) · QbitAI

Next · How companies use AIA 10-person game startup lets AI agents do most of the work →

✕Sun 1 Nov ‹›

How companies use AI · Sun 1 Nov

A 10-person game startup lets AI agents do most of the work

27times Spawn's founder had to decide himself, out of 11,000+ AI judge calls

Two young companies shared how they run with doing most of the day-to-day work: a game platform where agents write and fix the code, and a wealth manager where agents do the research and people make the money decisions.

Key points

Spawn (game-making platform, US): ten people work alongside an , permanent AI agents that each have a name, a memory and their own job, like the game engine, bug or finance. Founder Jacob Sansbury says they shipped 77,000 substantial code changes in 30 days, about 108 an hour, and spent about RM12.6 million (US$3.1 million) on AI last month, about 30 times payroll. Only about 0.9% of 123,548 changes in a year had to be undone (Jacob Sansbury's post, via The AI Daily Brief and Simon Willison).
Spawn turned two years of the founder's past decisions into an . In its first 8 days it handled more than 11,000 questions: about 7,300 settled by matching an earlier human decision, about 2,200 sent back to the agent, and about 1,800 passed to a person. The founder only had to decide 27 times, and daily changes went from about 1,000 to more than 7,000. Bugs that players' own agents report get fixed automatically, which makes it close to .
Arca (wealth manager, US): agents do the background research for its financial advisers. Some jobs run for up to 10 days, so each client has their own memory. People make the final calls and handle every trade and money move. Fixed check for , and a improves the setup using advisers' feedback (Latent Space).
Lesson: both give agents ongoing ownership of real work plus a shared record of what happened, while people keep taste, direction and money decisions. Spawn's numbers are its own, and its founder says big companies will find this way of working hard to copy.

Why it matters: This is what AI-native work can look like when a company is built around agents from day one. Most firms can borrow the simpler ideas, like turning past decisions into rules and letting agents own a task end to end.

Sources (3)

How the Most AI Pilled Company in the World Works · The AI Daily Brief
How I Built the Most AI-Pilled Company in the World (notes) · Simon Willison
Arca: the AI-native wealth manager · Latent Space

Next · AI at workCompanies start measuring what AI actually does at work →

✕Sun 1 Nov ‹›

AI at work · Sun 1 Nov

Companies start measuring what AI actually does at work

10.0%of Japanese firms surveyed see clear results from generative AI

Using AI is easy. Proving it pays off is harder. A big Japanese survey, new Claude Enterprise reports and a robot maker's new unit all try to answer the same question: what is AI really adding?

Key points

Japan survey (INDUSTRIAL-X, 1,031 people, July 2026): 60.9% of companies use generative AI, but only 10.0% say they see clear results. Most gains are time saved (72.3%) and costs cut (41.4%). Only 10.7% saw more sales (ITmedia).
What made the difference: when bosses took part, 61.2% used AI across the whole company versus 16.2% when they didn't, and 24.3% saw clear results versus 6.6%. A dedicated team and company-wide training also showed up much more often in the firms that got results.
Claude Enterprise added smart reports that show each team's work, costs, outputs and blockers, and suggest things like turning repeated work into . Anthropic says they are not for judging staff. It also added short in-app surveys and shared team budgets with email alerts at 50%, 75%, 95% and 100% (ITmedia).
Robots too: AgiBot proposed 'Worken', a unit for useful robot work, counted as active robots times hours working on their own times an output ratio. At one theme park in China it went from one engineer per robot to one per three, and aims for one per ten by year-end. It has not said exactly how the ratio is set (Pandaily).

Why it matters: The gap between trying AI and getting results is mostly about people and measurement, not the model. Clear numbers and leaders who join in are what move a company from testing to real gains.

Sources (3)

Only 10% see clear results from generative AI: what separates the winners (Japanese) · ITmedia Enterprise
Claude Enterprise adds smart reports, surveys and pooled budgets (Japanese) · ITmedia Enterprise
AgiBot proposes Worken to measure robot work · Pandaily

Next · AI in ChinaApple Intelligence reaches China in December with Alibaba's Qwen →

✕Sun 1 Nov ‹›

AI in China · Sun 1 Nov

Apple Intelligence reaches China in December with Alibaba's Qwen

18Apple Intelligence features in the first China rollout

After two years of waiting on regulators, Apple's AI features are coming to iPhones in China, powered by Alibaba's Qwen model. Chinese labs also shared new agent and document tools.

Key points

Apple says the first Apple Intelligence features arrive with iOS 27.3 in December, after a from 4 November. About 18 features come first, including writing tools, live translation for calls and messages, and visual intelligence through the camera (South China Morning Post).
DeepSeek published a design for a 'super-agent' that stays online and runs many jobs at once. A Router talks to the person and hands out work, Workers do the jobs in separate sessions, and a restarts things that get stuck. It notes people can only follow 3 to 5 sessions at a time. It is a test design with a on GitHub, not a product (IThome).
Xiaomi open-sourced Xiaomi--0, a small 0.8-billion-parameter model that reads documents, tables and receipts and returns neat data. Xiaomi says it scores 96.83 on the OmniDocBench test, first among models under 1 billion . These are Xiaomi's own numbers (Pandaily).
Not confirmed: code on DeepSeek's website points to a cybersecurity model called 'Cyber' with ID checks for users. DeepSeek has not announced it (Pandaily).

Why it matters: Apple needed a local partner to sell AI features in its second-biggest market, and Qwen got the job. Chinese labs keep shipping smaller, cheaper and more practical tools that anyone can try.

Sources (4)

Apple Intelligence to roll out on iPhones in China in December with Alibaba backing · South China Morning Post
DeepSeek publishes Super-agent design exploration (Chinese) · IThome
Xiaomi open-sources Xiaomi-OCR-0, a 0.8B document model · Pandaily
DeepSeek website code points to Cyber, a cybersecurity model in limited testing · Pandaily

Next · Southeast AsiaSingapore's NTU signs AI deal with Google; Vietnam eyes China help →

✕Sun 1 Nov ‹›

Southeast Asia · Sun 1 Nov

Singapore's NTU signs AI deal with Google; Vietnam eyes China help

40%of NTU undergraduate courses to include AI by 2030

A top Singapore university is putting Google's AI into its teaching and research, with a focus on Southeast Asian languages, and Vietnam is talking with China about working together on AI.

Key points

Singapore: NTU and Google signed a (MOU), an agreement to work together, on 30 October. It covers AI across the campus, regional-language models, science research and AI safety. NTU wants AI in 40% of undergraduate courses by 2030 and will look at linking Gemini Enterprise to its NTU Learn system (TechNode Global, The Star).
With AI Singapore, they plan to collect and datasets that capture how Southeast Asian languages really work, and build for education and healthcare on Google's open Gemma and MedGemma models.
Vietnam: at Vietnam International Digital Week, a Chinese ministry official offered joint research on large language models and open-source AI. Vietnam wants to be a leading AI centre in Southeast Asia by 2030 and among the world's top 10 by 2045 (TechNode Global).
Both are plans and promises for now, not results. No new Malaysian AI news turned up in our sources since yesterday's issue.

Why it matters: Models that understand local languages and culture matter for everyone in the region, including Malaysia. Watch for the open datasets, since those can be reused by others.

Sources (3)

NTU and Google deepen AI partnership · TechNode Global
NTU Singapore and Google deepen AI collaboration · The Star
China offers joint AI research with Vietnam as Hanoi sets 2045 target · TechNode Global

Today

Tuesday 3 Nov

Monday 2 Nov

Sunday 1 Nov