AI Weekly Malaysia

TSMC reported September net revenue of NT$511.86 billion (US$16.03 billion), up 54.6% year over year but down 0.6% from August; Q3 revenue was about NT$1.49 trillion. The company also committed in September to using ASML’s High NA EUV lithography machines, joining Samsung, and is scheduled to report Q3 earnings next week. CNBC corrected that September sales did not hit a record high.

OpenAI has rolled out Intelligent UI to 1.2 billion weekly ChatGPT users, replacing long text blocks with interactive graphics, maps, forms, and mini-tools generated by a streamable component library and inline compiler. It is not a new model: free and Go users get GPT-6 Luna, Plus/Pro/Business/Enterprise get GPT-6 Sol, and Pro keeps GPT-6 Astra, with Malaysia pricing unchanged at RM24, RM99.90, and RM420 per month. OpenAI claims interleaved thinking cuts web search wait times by up to 44%, but the article frames this as mainly a UI redesign over existing GPT-6 engines.

The npm package `tensorlake` (a TypeScript SDK for Tensorlake apps, sandboxes, and cloud services) was compromised in a ChainDrop / Shai-Hulud supply-chain attack; malicious version 0.5.144 has been removed from npm. Its preinstall hook launches `package/lib/setup.mjs`, which loads an obfuscated Bun-based worm (`package/lib/Math_Symbol.js`) that steals npm/GitHub/AWS/Vault/Kubernetes/SSH credentials, `.env` files, crypto wallets, messaging data, and config/MCP files for Claude, Cursor, Kiro, Windsurf, and Zed. It also drops HackBrowserData, persists on hosts, republishes victim-associated packages with Sigstore provenance, may plant GitHub Actions workflows, and resolves C2 via an Ethereum contract with GitHub as fallback.

In a 4-part Mastodon thread on mathstodon.xyz posted 2026-10-06, Terence Tao argues that in traditional mathematics ("Math 1.0") a breakthrough proof of a long-standing open conjecture triggers a whole follow-on ecosystem — invited talks, workshops, new collaborations, follow-up problems, and eventual digestion into textbooks and lecture notes. He then claims that responsibly used AI tools can support not just "the initial generation of a solution" but "all of the valuable" follow-up work, which he says a "Math 2.0" value system should measure more holistically. The Hacker News submission (2026-10-08) drew 338 points and 316 comments; the supplied text cuts off mid-sentence in part 2, so the rest of the argument and the thread replies are not available here.

Crystal Jade closed four more Singapore outlets on Oct 6—three Crystal Jade La Mian Xiao Long Bao branches and one Crystal Jade GO—leaving nine restaurants. Receivers Kroll, appointed Sep 29 for the Singapore and Hong Kong holding companies, also confirmed Crystal Jade Food Manufacturing (the central kitchen) ceased operations Oct 5 outside the receivership. About 120 workers received termination notices, while roughly 210 employees remain and Kroll plans to sell the business as a going concern.

Proton has officially revealed the 2026 X50 Sport Edition, its third SE release for the X-series SUV line after the 2025 X50 SE and 2026 X70 MC3, following a June 2026 preview at KLIMS. The changes are purely cosmetic — an all-black exterior with blacked-out chrome trim, 18-inch black alloy wheels, an SE bodykit with black/red accents, SE badging front and rear, and an all-black interior with red stitching, SE-embroidered front seats, metal scuff plates, SE carpets and TPE floor mats. No performance upgrade and no price yet; the X50 Flagship it is based on currently sells for RM113,300, and bookings opened on 8 October 2026 ahead of an official rollout later that month.

Tom's Hardware published a Prime Day roundup of PC maintenance tools, split into precision kits, screw kits, flashlights, dusters/cleaners, and thermal pastes. The supplied text is mostly site navigation and membership prompts; it does not name specific products, prices, or discounts. Published 2026-10-08, it is a consumer hardware deals piece, not AI/ML or developer tooling news.

Samsung Electronics reported preliminary third-quarter operating profit of 107.40 trillion won ($80.2 billion), the first time it has crossed 100 trillion won, up 782% year over year. Revenue came in at roughly 195 trillion won, up nearly 127% from the same quarter last year, after a Q2 that already set records at 171.5 trillion won revenue and 89.5 trillion won operating profit. The company attributes the surge to demand for memory used in AI infrastructure; detailed results are due later this month.

At a Windows event in San Francisco on 7 October 2026, Microsoft announced a Surface Ultra laptop built with Nvidia AI chips, with Jensen Huang and Satya Nadella appearing together on stage. Microsoft is pitching the device around a "hybrid intelligence" idea — more AI compute running locally on the machine instead of in cloud data centres. Malay Mail frames the launch as a bet that buyers will pay over RM10,000 for it.

A developer writes after a month of heavy use across a dozen projects that DeepSeek 4.1 Flash is subjectively indistinguishable from Opus in conversations, work, and speed, and that a $10/month OpenCode Go subscription makes it effectively unlimited. The post cites a roughly 437x KV cache reduction versus DeepSeek V1, expected session costs rarely above $1, and examples like $0.003 instead of $1 to reorganize desktop files, with Opus 5.5 sometimes used for final code review. The Hacker News discussion has 266 points and 224 comments.

The fetched page contains only Ars Technica's cookie-consent and privacy-center boilerplate (essential, social media, targeted, performance, functional, and audience-measurement cookies, plus a note that only GDPR-country and certain US-state residents can opt out via the Consent Management Platform). None of the actual article body about Microsoft's AI-friendly hardware or Windows changes is present — no product names, specs, prices, dates, or feature details appear in the text. The only verifiable facts are the headline, the URL, and the published timestamp of 2026-10-08.

Tom's Hardware flags the TeamGroup G50 EVO 4TB PCIe 4.0 SSD as the cheapest 4TB drive it has seen for Prime Day, at $399 — roughly 10 cents per gigabyte. The article describes it as a 'speedy' PCIe 4.0 drive at 'cut-throat pricing,' but the excerpt contains no benchmark numbers, no read/write figures, and no comparison table beyond the price-per-gigabyte claim. It is a deal post, not a review or measurement.

Lowyat.NET reports that Tabung Haji temporarily suspended its services after detecting suspicious cyber activity, in a post dated 8 October 2026. The excerpt contains only the headline and site navigation, so there is no information on which services were taken offline, how long the suspension lasted, how many users were affected, or whether data was accessed. Treat it as an unverified-until-confirmed incident notice, not a technical write-up.

Claude Dashboards and Claude Motion are now in beta. Docs, Slides, and Design are out of beta and on every Claude plan, including Free.

Bradley Axen, Head of AI Capabilities at Block, on how his team evaluates new large language models, how Claude Fable 5 orchestrates company-wide code migrations, and why safeguards are a critical part of deploying frontier intelligence responsibly, at scale.

A reference implementation walkthrough for a scheduled Claude Managed Agents automation, with the config files and six rules for avoiding common failure modes.

The post describes using Hugging Face's ML Intern to build a 0.8B prompt rewriter distilled from Qwen-Image 2.1's 9B rewriter, trained on 8,797 labeled example requests for about USD 16 total compute. The small model runs on CPU, returns valid output 99.7% of the time, and uses roughly a quarter of the teacher's tokens; the author then made five more models the same way via HuggingChat with ML-intern. The workflow stresses explicit prompts: a zero-shot baseline before training, a 50-step smoke test that checks saved weights changed, and spending caps such as USD 12 before approval.

Anthropic announced the Anthropic Cyber Mission on Oct 8, 2026, a long-term effort to support security defenders, starting with two programs: the Critical Infrastructure Defense Program (CIDP), which pairs frontier models, on-site engineers and threat research with operational-technology defenders in power grids, water systems and transportation, and OSS Scanner, which gives open-source projects free regular security scans from Anthropic's strongest models. Anthropic also says it merged Project Glasswing into an expanded Cyber Verification Program earlier that week, giving more defenders access to its most capable models. The post states that Project Glasswing partners found many vulnerabilities but that 'we haven't yet achieved a sufficient reduction in cyber risk' because verifying, prioritizing and fixing findings remains the hard part.

Anthropic committed $150 million over three years to the Genesis Mission, a US federal initiative, to put Claude into more than 15 agencies including NASA, NIH, and the National Science Foundation. The commitment covers Claude, Claude Code, and API credits for several hundred Genesis Mission research projects, plus training and onboarding, with stated focus areas of fusion energy and quantum computing. It was announced at the White House OSTP's "Science: A New Golden Age Summit" in Washington, DC, and follows earlier Anthropic science efforts: the Claude Science workbench, 10,000 free or discounted academic seats, and a research preview of the Model Hardware Standard for AI agents operating lab instruments.

Anthropic published its 2026 Usage Policy update on Oct 8, 2026, taking effect Nov 12, which mostly clarifies existing rules but adds controls for cases where Claude autonomously takes physical actions and tightens requirements for high-risk health and finance use cases. It consolidates disinformation, election, fraud and privacy restrictions into a new section, 'Do Not Engage in Deceptive Campaigns or Artificial Activity,' covering fake accounts, obscured authorship and infrastructure for influence operations, and renames the elections section 'Do Not Undermine Democratic Processes.' The blanket prohibition on personalized vote and campaign targeting is removed because it also blocked legitimate civic work such as nonprofits writing voter information in other languages or election officials sending ballot cure notices; the excerpt cuts off mid-sentence while describing the original rationale.

OpenAI says it banned two influence operations over the past two and a half years — one Russia-origin, one Iran-origin — that combined its models with conventional tactics to build "false front" entities. The Iranian operation ran seven "journalist" personas that pitched long-form articles to small and medium outlets worldwide and also mass-generated social media comments on US-Iran war topics; the Russian operation co-opted unwitting people in Latin America to run a ground-level "think tank" and produced fake "leaked" documents and audio scripts. OpenAI rates the Russia-origin operation as Category 5 on the IO Breakout Scale, which it calls a first, and notes both operations leaned on AI heavily to draft internal reports while using questionable methodologies to exaggerate their own effectiveness.

Mecka AI raised a $60M Series B led by Sequoia, with Nvidia, Microsoft's M12, and others participating. Founded in 2024, it pays people to record everyday tasks like making coffee or fixing cars while wearing body sensors and using smartphones, then analyzes that human motion data to train humanoid and other robots. TechCrunch previously reported the round was nearing a $500M valuation; competitors include XDOF, reportedly in talks for a Series B at a $1.2B valuation, and LLM data platforms Scale AI and Micro1 expanding into robotics.

Tom's Hardware flagged a rare discount on the WD Black SN770M 2TB, an M.2 2230 NVMe SSD aimed at Steam Deck and ROG Ally storage upgrades, at $0.13 per GB (roughly $260 for the 2TB drive, though the article text itself never states the total price). The piece calls it a return to near-launch pricing, which is the only hard number in the text.

Simon Willison's weblog republishes a quote from Ben Affleck describing how machine learning has long been part of visual effects work: he says he can write 'pretty shitty Python scripts' and explains using convolutional neural networks on tensors — the numeric encoding of a frame's pixels including batch number, frame number, and RGB values — for edge detection and feature extraction so green-screen elements can be replaced. He frames CNNs as precursors to transformers, which he describes as 'just much more computation simultaneously.'

Endeavor Catalyst closed its fifth fund at $320 million, lifting the firm's total assets under management past $850 million. It only invests alongside a lead investor in rounds of at least $5 million, writing $1M–$3M checks capped at 10% of the round, and plans 40–50 investments a year up to 150 companies from this fund. The catch: founders must already be inside Endeavor's network, which screened over 10,000 candidates last year and admitted 88, out of a network of 3,100+ entrepreneurs across 50+ countries.