AI Weekly Malaysia

Relay, an AI workflow automation startup launched in 2021 as a 'new Zapier,' is shutting down. Paying customers lose access September 14 (free users already lost it August 15). Founder and CEO Jacob Bank, who previously spent 6+ years at Google after his scheduling startup Timeful was acquired, is rejoining Google as VP of Product for Chrome, leading product and developer relations, with plans to integrate AI agents into the browser.

Why: If you were a Relay customer, you need to migrate your automation workflows before September 14. For everyone else, Bank's move signals Google is serious about embedding AI agents directly into Chrome — a browser with billions of users — which could reshape how developers build agent-driven experiences and how end users interact with web-based automation. The shutdown is also a reminder that VC-backed AI automation startups can disappear fast; don't build critical business processes on a single young platform without an exit plan.

Nvidia will provide credit and compute for a new OpenAI data center in Ohio, backing $105 billion in financing. SB Energy will build and manage the facility at the PORTS-Pike Technology Campus in Pike City, Ohio, under a 20-year lease to OpenAI, with initial capacity of 4.25 GW and an option for 3.75 GW more, coming online in phases starting 2028.

Why: Nvidia is moving from chip vendor to compute infrastructure financier, tying GPU supply directly to data center capacity for its largest customers. If you build on OpenAI APIs or depend on GPU availability, this signals that long-horizon compute capacity is being locked up by frontier labs through multi-decade deals—plan for a market where independent access to frontier-scale compute may get harder and more expensive, not cheaper.

404 Media used an AirTag hidden in a ~1,000-book order placed by an anonymous, price-insensitive buyer on Biblio to trace where bulk rare books are going. The shipment ended at Amazon's LAS8 facility in northeast Las Vegas, specifically the VGT3 section, where worker forum discussions confirm large volumes of books are destructively scanned—strongly suggesting Amazon is scanning books for AI training data.

Why: This is concrete evidence of the physical supply chain behind AI training data collection: companies are buying books in bulk and destroying them to scan content. If you ship AI products, this is relevant to copyright and licensing risk discussions—your models may be trained on data sourced through channels that could attract future litigation or regulatory action, especially as jurisdictions tighten training-data provenance rules.

Nvidia is investing $1.5B in SB Energy and extending up to $105B in credit to build OpenAI's Ports-Pike data center near Cincinnati, Ohio, securing itself as sole compute infrastructure supplier for a facility scaling from 4.25 to 8 gigawatts. The site includes a planned $33B, 9.2 GW natural gas power plant on former US uranium-enrichment land, with natural gas plant construction costs having risen 66% in two years.

Why: This signals that frontier AI compute is being vertically locked into Nvidia-supplied US mega-sites at unprecedented scale and cost, meaning cloud GPU pricing and availability outside the US will likely remain tight for years. Builders relying on GPU compute should factor continued US-centric infrastructure concentration into their architecture and cost assumptions, and consider whether multi-cloud or regional providers (including SEA options) offer better long-term economics than chasing US-based capacity.

America's largest electricity grid operator is proposing that new data centers consuming 50MW or more must bring their own on-site power generation or face being first in line for blackouts during supply shortages. The rule targets the massive power demands of AI-driven data center expansion and shifts the burden of grid reliability onto the largest new consumers.

Why: If you're planning AI infrastructure or choosing cloud regions, this signals that grid-constrained markets may increasingly deprioritize new data center loads — meaning higher costs, forced on-site generation requirements, or reliability risks for capacity you depend on. Malaysia's Johor data center boom could face similar regulatory pressure as grid capacity tightens, so builders should factor power availability into long-term cloud and colocation decisions now.

A threat actor called 'TheHatman' is selling millions of employee records allegedly exfiltrated from Microsoft Azure environments of nine major companies, including McDonald's (1.7M records), TCS (800K), Vodafone (425K), and HCL (250K). Hudson Rock assessed the data as 'highly likely authentic,' noting it contains phone numbers, physical addresses, job titles, reporting structures, group memberships, and—critically—identities of Global Administrator accounts, making it a phishing and privilege-escalation shortlist.

Why: If you run Azure AD/Entra ID, this illustrates what a directory export actually leaks beyond email: org charts, service account details, and who holds Global Admin. Audit your Entra ID for infostealer-compromised credentials, enforce phishing-resistant MFA on all admin accounts, and review third-party app permissions—Hudson Rock specifically flagged overly permissive third-party apps as a likely vector. The data exposure is severe even without passwords, because knowing your Global Admins by name is enough for targeted attacks.

A Japanese repair shop offers GDDR6 VRAM upgrades at $25 per GB, modding an RTX 2080 Ti from 11GB to 22GB for $282 total. This doubles the card's VRAM at a fraction of the cost of buying a higher-VRAM GPU, positioned as a budget option for AI workloads during a GPU memory shortage.

Why: If you're running local LLM inference and hitting VRAM walls, a $282 VRAM mod on an older card like the RTX 2080 Ti could let you load larger models locally without spending $1,500+ on a new high-VRAM GPU. The $25/GB price point is worth benchmarking against the cost of cloud GPU rentals for your actual workload.

Black Hat and DEF CON 2026 in Las Vegas were dominated by AI agent security concerns, with discussions centering on rogue agents escaping sandboxes and threats to critical infrastructure. Government and former government leaders expressed worry, and even conversations about US water infrastructure attacks looped back to AI.

Why: If you're shipping AI agents, the security community is now treating sandbox escapes as a real and expected failure mode rather than a hypothetical. This should push you to treat agent containment, permission scoping, and output validation as first-class engineering problems before deploying agents that can take actions on infrastructure or customer systems.

Linux 7.2 is released with Torvalds acknowledging that AI coding tools have increased contribution volume to the point where releases are messier but delaying is no longer viable — 'If I delayed releases for that reason we'd probably never have a release at all.' The kernel adds Cache Aware Scheduling for manycore chips like AMD EPYC 5 and Intel Xeon 6, Apple M3 support, and next-gen AMD/Intel/Nvidia chip readiness, while dropping AppleTalk and old ISA/PCMCIA ARCNet adapter support.

Why: Torvalds explicitly attributing larger, messier release cycles to AI-assisted contributions is a concrete signal that AI coding tools are now reshaping even the most rigorous open-source development process — if you maintain or contribute to shared codebases, expect review burden and revert frequency to rise. On the infrastructure side, Cache Aware Scheduling on EPYC 5 and Xeon 6 means workloads on those chips should see better cache utilization without application changes, worth benchmarking if you run dense multi-core workloads.

Singapore's GDP grew 5.9% YoY in Q2 2026, driven largely by AI-related demand boosting manufacturing (electronics output +33.8%, precision engineering +19.3%) and finance (+6.2%), while retail (+1%), accommodation (+2.2%), and F&B (-1.5%) lagged. Productivity gains are concentrated in outward-oriented industries (+6.9%) while domestically-oriented sectors saw productivity fall 0.1%, creating a widening gap between AI-adopting and non-adopting sectors.

Why: For Malaysian founders and builders, this is a leading indicator of how AI-driven growth concentrates in semiconductor, cloud, and finance supply chains while leaving consumer-facing businesses behind. Companies positioned in AI infrastructure, wholesale trade, or fintech should expect tailwinds; those in retail, F&B, or professional services face stagnation unless they find AI-leveraged productivity plays. MTI's finding that initial AI employment gains accrue to higher-earning and skilled foreign workers suggests hiring strategies should prioritize AI-adjacent roles now.

Chinese AI company Zhipu launched GLM-5.3, claiming it outperforms Fable 5 and GPT-5.6 Sol on the CyberGym benchmark for vulnerability discovery. The model reportedly found 2,436 vulnerabilities across 269 real-world projects, including 1,097 medium-to-high severity issues spanning kernels, browser engines, and network protocols—some dating back roughly 40 years. However, GLM-5.3 performed worse than Western models on other security and coding benchmarks.

Why: If you ship code, AI-assisted vulnerability discovery is becoming a commodity capability across multiple model providers, not just US labs. The 2,436-bug finding across 269 projects suggests you should evaluate AI bug-finding tools against your own codebase now rather than waiting for a single dominant provider. The fact that GLM-5.3 underperforms on general coding benchmarks but excels at exploitation-chain reasoning means bug-finding may be a distinct capability axis worth benchmarking separately when choosing models.

John Gruber reports that Anthropic will watermark all Claude-generated text worldwide via steganography — manipulating word/token choice at inference time to leave probabilistically detectable fingerprints — to comply with EU regulation. He argues this directly contradicts Anthropic's own support document, which claims the watermark is 'imperceptible' and 'doesn't change the meaning, quality, or readability' of output, when the technique inherently alters word selection.

Why: If you ship products or pipelines on Claude APIs, your output text will soon be subtly modified in word choice as part of watermarking — not via invisible characters, but by changing which tokens are selected. This could affect output consistency, fine-tuned style expectations, or downstream NLP processing. Evaluate whether this matters for your use case before it rolls out, especially if you rely on deterministic or stylistically controlled outputs.

Nvidia has dramatically scaled back the $250 billion data center financing guarantee it was reportedly willing to provide to OpenAI, according to a WSJ report surfaced via Reuters. The reduction signals a significant pullback in Nvidia's willingness to underwrite OpenAI's compute infrastructure buildout.

Why: If Nvidia is retreating from bankrolling OpenAI's compute expansion, founders and AI teams building on OpenAI APIs should consider that OpenAI's future compute capacity and pricing may be less predictable than assumed. This could also signal broader caution about AI infrastructure spending, which affects anyone budgeting for GPU-dependent workloads or planning around OpenAI's roadmap.

A reflection arguing that AI agent harnesses have crossed the 'can it be done' threshold but that software engineering fundamentals—how code fits together, API design, testability—matter more than ever. The author notes open-weight models are closing the capability gap fast on local hardware, and shares a concrete practice: prompting agents with 'develop with red/green TDD' to get testable output, though code seams above that level remain subjective art.

Why: If you're using agentic coding tools, the bottleneck has shifted from 'can the agent write code' to 'can you architect the seams between modules and specify tests.' Try explicitly prompting for TDD and treat the agent as a fast welder, not an architect—your judgment on API boundaries and integration is what keeps the output shippable.

David Horn argues that software engineers repeatedly reinvent concepts from older disciplines—statistics became 'data science,' crypto speedran finance history, Silicon Valley reinvented bus stops—without studying the originals. He sees the same pattern repeating with AI agents, where engineers are intuiting that managing multiple agents resembles managing engineering teams, but are not consulting existing management literature.

Why: If you're building AI agent workflows, the article's core point is that the leadership/management intuition you're developing about orchestrating agents is not new—decades of management science already cover delegation, feedback loops, and team coordination. Before designing your own agent orchestration patterns from scratch, check whether established management frameworks map onto your problem.

Simon Willison built CORS Chat, a browser-based tool for chatting directly with any OpenAI Responses-compatible API endpoint that supports CORS headers. He used it to test Qwen 3.8 27B running in LM Studio on an M5 MacBook Pro and an NVIDIA DGX Spark, with conversations persisted locally and exportable as JSON. It also progressively renders SVG images while tokens are still streaming.

Why: If you run local models via LM Studio (with the --cors flag) or use OpenRouter, this gives you a zero-backend, browser-only chat UI for testing endpoints—useful for quick model evaluation without spinning up a separate frontend. The progressive SVG rendering detail is worth noting if you build streaming UIs.

Apple denied developer Sean Byrne access to App Store Connect after an automated match against a U.S. government Consolidated Screening List entry for a 'Sean Byrne' in County Sligo, Ireland — an entity that appears to be a fabricated employee from a 2009 aircraft-parts export prosecution. Despite Byrne providing his passport, driver's license, and pointing out he has no connection to the listed address or company, Apple has not responded to his appeal.

Why: If you ship apps through App Store Connect, you are subject to automated sanctions screening with no apparent human escalation path — a false positive can lock you out of your livelihood indefinitely. Malaysian and non-U.S. developers should be aware that this screening covers international lists, not just U.S. persons, and that Apple's process offers no documented timeline or appeal mechanism once a match is flagged.

Anthropic announced a text watermarking scheme for Claude that modifies inconsequential word choices (e.g., picking 'cold' over 'gray' when meaning is unchanged) to comply with the EU AI Act, based on Google DeepMind's SynthID-Text technique. The watermark uses a separate randomness source detectable with a digital key, and other AI model makers are expected to deploy similar approaches.

Why: If you ship LLM-powered products to EU users, watermarking will likely become a compliance requirement you can't opt out of — and since it alters token sampling, it could subtly affect output quality or determinism in ways that break evals or downstream pipelines. Builders using Claude (or eventually other models) should test whether watermarked outputs behave differently in their specific use cases, especially for tasks sensitive to word choice like summarization or classification.

Apple submitted a court-ordered proposal to charge 15% commissions on purchases made via external links inside iOS apps, with 5% for small business developers, 10% for certain partner programs, and 10% for subscription renewals. The Supreme Court rejected Apple's bid to delay, forcing the disclosure after Apple had previously imposed a 27% external-link commission that the court found problematic. Apple compared its rates to Google Play's 20% standard, 15% special program, and 10% subscription renewal link-out fees.

Why: If you ship an iOS app and route payments through external links, these proposed rates determine your unit economics: 5% if you qualify as a small business developer versus 15% standard. Founders should model both scenarios now and check whether their app qualifies for the small business or partner program tiers before finalizing pricing.

Hyperscalers (Amazon, Google, Meta, Microsoft) are betting heavily on natural gas to power AI data centers, with Meta planning a 7.5GW plant in Louisiana, Amazon 7.6GW in Texas, and Microsoft and Google each building gigawatt-scale gas plants in Texas. Energy research firm Noreva forecasts natural gas prices could triple above $10/MMBtu in certain U.S. hubs (from ~$2-4.50 today), as hyperscaler demand collides with declining supply growth and rising LNG exports.

Why: If fuel costs double or triple, cloud compute pricing for AI workloads could rise materially since fuel is roughly half the cost of electricity from large gas plants. Founders and developers running GPU-heavy workloads on AWS, Azure, or GCP should model scenarios where cloud inference and training costs increase, and consider cost-optimization strategies like spot instances, model distillation, or multi-cloud arbitrage before locking into long-term cloud commitments.

Cloudflare announced new Cloudflare One capabilities to detect and control MCP (Model Context Protocol) traffic on corporate networks. The core problem: employees can connect AI agents (Claude Code, Codex, Cursor, VS Code) to arbitrary MCP servers with a single line of config, and the resulting HTTPS traffic has no distinguishing hostname or path pattern, making 'shadow MCP' usage hard to spot. Cloudflare Gateway now uses protocol signals to identify this traffic and enforce routing through approved MCP Server Portals.

Why: If you are shipping or using AI agents that connect to MCP servers, your existing permission models were designed for humans who pause on unexpected results and act at human speed. Agents act non-deterministically and can repeat a bad tool call thousands of times before anyone notices. You should decide now whether your team needs network-level visibility into which MCP servers agents are calling, especially if employees can self-configure connections without approval.

Grab PayLater can now be used at any physical merchant displaying a DuitNow QR code, not just Grab-specific QR standees. The feature is rolling out to a whitelisted group of users, who also earn up to 2x GrabCoins when paying via PayLater on a DuitNow QR code.

Why: If you build or sell into Malaysian retail payments, BNPL is now effectively universal at the point of sale via DuitNow QR—any merchant with the standard national QR code is a potential BNPL checkout. SaaS founders in POS, e-commerce, or fintech should evaluate whether their flows or merchant onboarding need to account for Grab PayLater as a payment rail alongside standard DuitNow QR processing.

OpenAI, Google DeepMind, Anthropic, and Chinese firms like Alibaba and ByteDance are expanding aggressively in Singapore, with fresh AI hires earning S$70,000–S$90,000 and PhD-level specialists pulling S$200,000–S$350,000+. OpenAI committed S$300M+ and opened its first Applied AI Lab outside the US with 200+ planned technical roles, while roughly 50 Chinese AI firms have set up Singapore operations since 2024—though the article questions how durable that commitment is, noting some may simply be acquiring a Singapore address for Western market access.

Why: Malaysian developers and AI engineers now have concrete salary benchmarks for the regional market one border away—use these numbers when negotiating remote or relocation offers to Singapore. Malaysian startups competing for ML talent should expect upward pressure on local compensation and plan retention strategies accordingly, since the talent pool is shared.

OpenAI launched 'Computer History,' an opt-in feature for the ChatGPT macOS desktop app that records clicks, typing, keyboard shortcuts, and app switches to build timeline-based memories for ChatGPT. It replaces the earlier screenshot-based 'Chronicle' with input-event capture, storing files unencrypted locally for 48+ hours, with OpenAI warning other macOS user-level programs can access them. Available to Pro, Business, and Enterprise users (not EEA, UK, Switzerland, API, or Bedrock users).

Why: If you use the ChatGPT macOS desktop app on a Pro or Enterprise plan, this feature is off by default but can be enabled by admins or individuals—know what it captures before turning it on, since unencrypted local files containing your keystrokes and clicks are accessible to other processes running as your macOS user. Builders shipping agent tools should note this as another prompt-injection attack surface where sensitive user activity flows into LLM context.

Researchers at Reco have tracked an attacker dubbed "City-Forum" spending over a year harvesting data from over-permissioned Salesforce and ServiceNow guest portals worldwide, targeting telecoms, banks, cybersecurity firms, and public sector bodies since at least March 2025. The attacker used custom tooling against Salesforce Lightning Web Runtime sites via the UI API's GraphQL layer and a little-documented ServiceNow Service Portal search endpoint, exploiting misconfigured guest permissions rather than platform vulnerabilities. The busiest target logged over 560,000 enumeration events from the attacker's IP.

Why: If you ship customer or partner portals on Salesforce or ServiceNow, audit your guest user permissions and disable self-registration where possible—this campaign proves over-permissioned guest accounts are being actively and systematically raided right now. The attack vector is configuration, not a CVE, so no patch will save you; only tightening guest access and reviewing what records guests can read will.