AI Weekly Malaysia

China's ChangXin Memory Technologies (CXMT) said over the weekend it made progress on a new memory architecture that packs more data into a smaller space, and memory stocks fell sharply: SK Hynix closed 10% lower, Samsung dropped 7.4% and Micron was down 3.7% in premarket trading. CXMT IPO'd in July on Shanghai's STAR Market, soaring nearly 466% on debut to become the most valuable China-listed company at the time; its shares now sit about 4% below the 49 yuan IPO-day price. The report gives no yield, capacity, product timeline or pricing numbers for the new design.

Licube is a startup that has licensed lithium-6 purification technology from Hirosaki University in Japan, betting that fusion plants will need the isotope to breed their own tritium fuel on site. It pitches that lithium-6 is only about 7.5% of natural lithium and that separation capacity is scarce today, with co-founder and CEO Yohei Kiguchi telling TechCrunch it "should be a $50 billion market by 2040" if fusion comes online. Licube is one of 20 startups presenting Tuesday at TechCrunch Disrupt's Startup Battlefield in San Francisco.

The fetched page for this Ars Technica item (published 2026-10-13) contains no article body at all — only Condé Nast's cookie/consent-preference boilerplate listing opt-out rights for US states. The only substantive information is the title itself: a taxpayer-funded Trump commercial railing against his political enemies reportedly kept airing on TV the day after a judge blocked it. No figures, dates, court name, dollar amounts, or quotes from the ruling are present in the text.

Apple announced HomeView, a 7-inch screen smart home hub that can be wall-mounted or set on a table, starting at $299 and shipping Nov. 13. It recognizes household members by face to show personalised information and is positioned as the showcase device for new Siri AI capabilities. Apple also refreshed Apple TV 4K at $299 with a redesigned remote that plays a sound when lost, and HomePod Mini in new colours at $149, up from $129 — the second launch event under CEO John Ternus, who took over from Tim Cook in September.

Guitoro is a free browser-based virtual guitar (Show HN, 489 points / 129 comments) that you play with your keyboard: left hand holds one of 15 chords on letter keys while the right hand strums with H and J, plus arpeggios, single strings, chucks and palm mutes. It ships 4 instruments (steel-string acoustic, nylon/classic, electric, 4-string bass), 51 songs, 56 strum/picking patterns and 15 drum styles, plays along at speeds down to 20%, and every note is a recording of a real instrument with no download and no account. Touch and mouse are first-class too: chord pads plus up to five-finger swipes on mobile, and a fast mouse swipe plays louder while a slow drag rolls notes one by one.

Dell Technologies CEO Michael Dell told CNBC's David Faber that AI models "have to have safeguards and controls," comparing the technology to fire that "can heat your home" but "can also burn down your whole village." He said "there's a role for government to play" and that AI builders should set "bounded permissions," but declined to give specifics on what that government role would look like, saying he's "not a policymaker." The article also notes that Anthropic's Dario Amodei and OpenAI's Sam Altman have backed efforts to pace frontier AI development, and that opposition to AI data centers has become a key issue ahead of the 2026 midterm elections.

Niobia, a San Francisco-based startup and TechCrunch Disrupt 2026 Battlefield finalist, sells AI software for battery and chip manufacturing lines: AI agents run on self-contained GPUs mounted directly on production equipment, flag pinhole-sized defects like scratches, and feed a chatbot that engineers query to find root causes — aiming for answers in minutes instead of days. Co-founders Gaurav Jha and Harveen Kaur both hold chemistry PhDs and have worked at chipmakers and battery manufacturers. The piece gives no pricing, no named customers, no funding figures, and no accuracy numbers.

Mind Robotics, the industrial AI and robotics venture Rivian created last year, has hired Doug Field as president, announced onstage at TechCrunch Disrupt 2026 (Oct 13-15, San Francisco). Field, who previously worked on EVs at Ford and Tesla, Apple's car project, and was employee No. 1 at Segway, will run day-to-day product, engineering, and operations, while Rivian CEO RJ Scaringe stays involved. The startup has raised more than $1 billion from Accel, Andreessen Horowitz, Eclipse, Kleiner Perkins, and the venture arms of Volkswagen and Salesforce, and is building purpose-built, non-humanoid robots running foundation models for industrial and manufacturing tasks, with Rivian as its first customer and live training environment.

Mind Robotics, the industrial AI and robotics venture spun out of Rivian last year, has hired Doug Field as president — announced onstage at TechCrunch Disrupt 2026 (October 13–15, San Francisco). Field's prior roles per the article: EVs at Ford and Tesla, Apple's car project, and employee No. 1 at Segway. CEO RJ Scaringe stays involved but Field takes day-to-day product, engineering and operations; the company has raised more than $1 billion from Accel, Andreessen Horowitz, Eclipse, Kleiner Perkins, and the venture arms of Volkswagen and Salesforce, and is building purpose-built (explicitly non-humanoid) robots running on foundation models, with Rivian as its first customer and a "live, high-volume manufacturing environment for model training and deployment."

Flox Intelligence, a startup led by Sara Nozkova after 13 years in transportation and autonomous vehicles, is building a wildlife-deterrent system for train tracks, roads and airports that deliberately avoids the usual playbook of sirens, lights, fences, falcons and culling — Nozkova's stated premise being that "the answer was not a louder deterrent." The cost problem it targets is well quantified in the piece: the FAA puts U.S. civil aviation wildlife strikes at about $272 million a year on average from 1990 to 2025, while SNCF says animals affect more than 8,000 French trains annually, and in Brittany 120 trains hit animals in 2023, disrupting 660 services and causing over 15,000 minutes of delays. The article supplies no funding, pricing, customer, deployment-scale or technical-architecture detail.

Ars Technica reports that Disney offered $20 million for Character.AI's chatbot technology, describing Character.AI as one of the first chatbot makers and tying it to a child suicide scandal. The article (published Oct 13, 2026, 27 comments) frames the move as "enemies-to-friends." The excerpt provided is mostly site navigation, so deal terms, structure, and Character.AI's response are not visible in the text.

The Hacker News item is headlined 'Anthropic Warns Small AI Models Could Soon Form Self-Propagating IoT Botnets,' published 13 October 2026. The retrieved text contains only the site's navigation, subscription prompts, and advertising boilerplate — no model names, size thresholds, capability evaluations, timelines, or quotes from Anthropic are present. There is no technical detail in this excerpt to summarize beyond the headline claim itself.

Ars Technica reports that Discovery+ subscription prices are rising by up to 20 percent under Skydance. The article body as provided contains only the site's cookie consent and privacy boilerplate — no pricing tiers, no effective dates, no affected regions, and no explanation from the company. The only concrete fact available is the headline figure of a hike of up to 20 percent.

Meta launched "Your Algorithm" on Facebook on October 13, 2026 (announced Tuesday), after first shipping it on Instagram. The feature lets users describe in natural language what they want to see more or less of in Feed and Reels, with Meta saying changes are reflected in recommendations in real time. Facebook's three-dot post menu was also updated so users can open "Your Algorithm" or tap "Interested"/"Not interested", and Meta says it may ask follow-up questions to refine feedback.

Longshot Space Technology, an Alameda, California startup co-founded by CEO Mike Grace and CTO Nato Saichek, is building a ground-based gun that fires cargo into space instead of using a first-stage rocket. It has raised about $20 million and is adapting 'multi-injection gun' designs — a horizontal steel pipe with high-pressure gas tanks attached at ports along its length that fire as the projectile travels down the barrel — originally built for World War II and 1960s research programs. The pitch, per Grace, is doing for that existing gun research what he says SpaceX did with NASA rocket research: scaling already de-risked technology rather than inventing something novel. The piece is a TechCrunch Disrupt 2026 session promo and the excerpt cuts off mid-explanation.

The Hacker News reports that Docker has fixed a critical flaw in Docker Hub that allowed one user to push images as another user, and that the bug had been present for roughly two years before the fix. The supplied article text contains only the headline and site navigation boilerplate — no CVE identifier, affected versions, patch version, disclosure timeline, or statement from Docker is included.

The extracted page content for this Ars Technica item contains nothing but the site's cookie/privacy consent notice — cookie categories (social media, essential, targeted, performance, functional, audience measurement) and a note that only GDPR-country and certain US-state residents can opt out via the Consent Management Platform. No article body, headline detail, car models, specs, or pricing made it into the text. The only usable metadata is the title, URL, source, and published timestamp of 2026-10-13.

The Hacker News reports that an attacker chained three zero-days in Ivanti Connect Secure to gain root access, and that only two of the three had been patched at the time of publication (13 Oct 2026). The excerpt provided is almost entirely site navigation and boilerplate: it contains no CVE identifiers, affected version numbers, IoCs, or vendor mitigation steps. Beyond the headline claim, there is not enough text here to summarise the technical chain or the patch status per flaw.

SAP disclosed a CVSS 10.0 severity vulnerability in its Cloud SDK for Python that allows OAuth2 credential theft, reported by The Hacker News on 13 October 2026. The supplied text is truncated at the headline, so the CVE identifier, affected SDK versions, attack preconditions, and patch guidance are not available in this source.

Pluto Security emerged from stealth with $42 million after starting sales in April, crossing $2 million in ARR with 30+ enterprise customers including Oscar Health, Workato, Sixt, and AlphaSense; its $8.5M seed was led by TLV Partners with Armis' founders investing. CEO and co-founder Shahar Bahat says a newly released AI tool can appear inside a company's environment within two hours of launch, and that roughly 60% of the risks Pluto detects across its customer base sit in tools vibe-coded by non-developers. Finance departments are the top source of those risks, followed by marketing and legal, with developers fourth; Pluto cites a Fortune 500 finance employee's vibe-coded customer/churn app left exposed to the internet.

LiquidAI released LFM2.5-24B-A2B, a 24B-parameter MoE reasoning model with only 2B active parameters per token, positioned for running agents on your own hardware. It reports 184 tok/s on an Apple M5 Max and up to 21K output tok/s on a single NVIDIA B300 GPU, and claims 48.09 on LiveClaw versus 38.19 for Qwen3.6-35B-A3B, a model with roughly 50% more total and active parameters. Versus its LFM2-24B-A2B predecessor it adds an explicit chain of thought, grows context from 32K to 131K tokens, doubles tokenizer vocabulary from 65K to 128K for non-Latin text, and scales pre-training from ~17T to ~33T tokens.

The supplied text contains no article content — only Ars Technica's cookie-consent and privacy-center boilerplate (GDPR-country and certain US-state opt-outs, essential/functional/targeted/performance cookie categories). The only substantive signal is the headline: 'Google wants to host your personal vibe-coded apps in the Play Store.' No program name, dates, eligibility rules, pricing, review requirements, or technical detail are present in the text, so nothing can be confirmed beyond the claim in the title.

Google Play added a feature letting users run their own AI-built ("vibe-coded") apps privately on Android devices under their standard Google account, without publishing them publicly to the Play Store. The apps are built with Google AI Studio, don't appear in Play search or listings, and are visible only to their creator, though they must still comply with existing Google Play policies. The feature launches initially in the U.S. for users 18 and up; Google says publishing to Play previously exposed apps to the store's roughly 3 billion users, which pushed many people to keep vibe-coded projects as web apps instead of native Android.

LNK Energies, an MIT-linked startup, claims it can convert existing Class 8 diesel trucks to run on hydrogen carried as a liquid organic hydrogen carrier (LOHC), with CTO Alexis Huynh telling TechCrunch the total cost of ownership is 13% cheaper than diesel. The core technical claim is reactor miniaturisation: LOHC reactors that are normally "about the size of a shipping container" have been shrunk "to the size of a mini fridge," small enough to ride on the truck instead of at a fuelling station. The item is a TechCrunch Disrupt 2026 promotion (published Oct 13, 2026) and the supplied text cuts off mid-sentence before the energy-saving explanation is finished.

Show HN project "How random are you?" (lab.specr.net/rigged-draw, credited on the page to Shaun Pedicini) turns a browser page into a live guessing game: you say digits 1-9 out loud as randomly as you can, and the house seals its guess in an envelope before each number, then opens it. It runs speech recognition fully in-browser via a ~31 MB one-time speech pack (audio never leaves the page), skips the first 10 numbers as warm-up, and scores your hit rate against 206 people from a psychology study while seeding its model from 418 people who each picked 200 numbers from 1 to 9. Ten small predictors run at once and it trusts whichever has been right lately, with the "numbers you haven't said in a while" predictor doing most of the work. The thread drew 213 points and 119 comments.