Report a problem

Blog

Writing about AI — how it works, where it helps, and what we learn building with it.

Latest posts

Bitget's official brand graphic: the Bitget wordmark and logo on a plain dark background, used by the exchange as the preview image for its own site.Bitget's Root-Cause Report Says the Attackers Came In Through Its Security Products. The Appliance That Watches Everything Can Also Reach EverythingBitget published its breach findings on September 30: a zero-day in a third-party security appliance gave attackers a foothold on August 31, three and a half weeks before $387.5 million left its wallets. The defensive control was not bypassed — it was the route in.SecurityThe GitHub social card for the Tencent-Hunyuan/Hy3 repository, showing the repository name and its description of Hy3 as a 295B-parameter Mixture-of-Experts model with 21B active parameters.Tencent Is Reported to Have Rented 100,000 Chips It Cannot Buy. Export Controls Regulate Where Compute Sits, Not Who Uses ItThe Financial Times reported on September 30 that Tencent has leased about 100,000 advanced AI chips from Oracle for roughly $7 billion over five years, hosted in Southeast Asia. Tencent cannot buy those chips; renting them offshore is legal, and the House bill written to close that gap has been sitting in the Senate since January.Artificial intelligenceGoogle's official Gemini 4 Argon key art, the wide title graphic published with the company's own announcement post.Google's First Frontier Model Since Gemini 3 Went to Cyber Defenders Before It Went to Customers. The Order Is the AnnouncementGoogle says Gemini 4 Argon leads or ties on 13 of 18 benchmarks. Nobody outside a vetted programme can check, because the model went first to cyber defenders and a US government review process and only later to the people paying for tokens.Artificial intelligenceThe Federal Trade Commission's own default social-sharing graphic, the branded agency image published on ftc.gov as the standard preview for FTC pages.The FTC's First Rogue-Agent Probe Names METR Alongside OpenAI and Anthropic. The Auditor Was Supposed to Be the Answer, Not a SubjectThe FTC confirmed on Wednesday that it is investigating OpenAI, Anthropic and other AI developers over the consumer risks of autonomous agents, using the 1914 FTC Act rather than any new AI law. The third name on the list is METR — the nonprofit evaluator the industry offered as its own check.Artificial intelligenceApple's official press image of the HomePod mini colour lineup: five HomePod mini speakers in a row on a plain background.Apple Is Reported to Enter the Smart Home on October 13. The Hub Is Not the Product — Siri IsBloomberg says Apple will announce a new home hub, a new HomePod mini and a new Apple TV on October 13. Apple has confirmed nothing — but the reported design gives away what the device is actually for, and it is not sound.AppleThe GitHub social card for the deepseek-ai/DeepGEMM-Ascend repository, showing the project name and its description as a matrix multiplication kernel library for Huawei Ascend NPUs.DeepSeek Open-Sourced Six Libraries for Huawei's Ascend Chips. The Software Layer Is the Part Export Controls Never TouchedDeepSeek published a TileLang backend for Huawei's Ascend 950 and five kernel and communication libraries on September 30, developed with Huawei and merged upstream next to the CUDA backend. Export policy restricted the chips; it never restricted CUDA, and CUDA was always the harder thing to replace.Artificial intelligenceOpenAI's official DevDay 2026 art card, the 16:9 title graphic published with the company's own DevDay recap post.OpenAI Announced More Than Twenty Things and Only One Was a Model. The Rest Is an Attempt to Become the Place Software LivesOpenAI's DevDay shipped a word processor, a slide editor, a shared drive, an identity provider, a plugin UI surface and an enterprise procurement channel. What it did not ship is a billing system — and an app store without a cash register is a subsidy, not a business.Artificial intelligenceAnthropic's official brand image, used as the sharing image on its newsroom: the Anthropic logo on a plain background.Anthropic's Leaked Prospectus Commits $518 Billion to Compute Against $4.6 Billion of 2025 Revenue. Most of It Cannot Be CancelledReuters and the Financial Times have reported from a leaked June draft of Anthropic's IPO prospectus: $4.6 billion of 2025 revenue, an $8.06 billion operating loss that is almost entirely compute, and $518 billion of cloud commitments that are roughly 80% non-cancellable. The headline loss is not the disclosure that matters.Artificial intelligenceThe flag of the United States Federal Bureau of Investigation: the FBI seal centred on a dark blue field, with the bureau's name in gold lettering.The FBI Made a Video for ShinyHunters' Remaining Members. The Arrest It Announced Happened Two Weeks AgoDutch police arrested an alleged ShinyHunters leader on 15 September; the FBI made it public this week alongside a video telling everyone still in the group to come forward. The Bureau put numbers on the crew for the first time — 140 organizations, $70 million — and the reason it is spending this much attention is the group's own decision to go after FBI personnel records instead of money.Security