Report a problem

Blog

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

Latest posts

The 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.A Second ShinyHunters Figure Is in Custody, and This One Is Walking the FBI Through His Own Devices. The Group's Channels Went Dark the Same DayReuters reports that Saif al-Din Khader — the hacker known as "Rey" — was detained in Jordan on Tuesday and is helping the FBI identify the rest of ShinyHunters. The detention rests on anonymous sources, but the group's communication channel went unreachable the same day and its leak site was gone by Wednesday.SecurityApple's press photograph of the iPhone 18 Pro colour lineup, the handsets laid out side by side in each finish with their rear camera systems facing the camera.Apple Has Finally Spoken on the iPhone 18 Pro Max AT&T Failure. The Phones Already Broken Have to Be ReplacedApple's first public statement on iPhone 18 Pro Max units losing service on AT&T names two updates meant to prevent the fault — and confirms that phones already stuck in SOS cannot be recovered in software and must be swapped for new hardware.AppleApple's own press photograph of the latest Mac mini and Mac Studio, the desktop Macs on which macOS grants or withholds Full Disk Access.Apple Says It Will Add New Controls to Full Disk Access Because of AI Agents. It Named No Version, No Date and No DeveloperApple told developers on October 2 that it will require "very explicit user action" before an app can be granted Full Disk Access, the one macOS permission that bypasses the system's per-item privacy prompts entirely. The note names AI agents as the reason — and promises a harder door rather than a smaller one.AppleOpenAI's official dark-toned cover graphic for its own report on the Hugging Face incident, published on openai.com.OpenAI Fired Three Safety Researchers for Sharing Information Outside the Company. One Was Its Own Contact for the Outside InvestigationOpenAI confirmed it dismissed three safety staff for mishandling sensitive information, without naming them, the outside group, or the material. Reporting identifies the recipient as METR and one of the three as OpenAI's own technical contact for METR's Hugging Face review.Artificial intelligenceApple's press photograph of the iPhone 18 Pro and iPhone 18 Pro Max standing side by side, angled to show the rear camera system.AT&T Is Warning iPhone 18 Pro Max Owners Their Phones May Not Reach 911. The Update It Tells Them to Install Documents a Different FixAT&T has begun texting iPhone 18 Pro Max customers that some phones cannot complete calls, texts or data "including to 911," and telling them to install iOS 27.0.1 immediately. That update's release notes describe a Face ID fix, and affected owners say it does not help.AppleApple's press image of the iPhone 18 Pro in burgundy, shown front and back side by side, the front displaying its lock screen.A Forensics Vendor Told Police It Can Keep a Seized iPhone Unlocked Indefinitely. The Video Making the Claim Is Eighteen Months OldA leaked law-enforcement training video reviewed by 404 Media shows Magnet Forensics marketing a way to hold a seized iPhone in its After First Unlock state through Apple's 72-hour Inactivity Reboot. The capability is an unverified vendor claim; the eighteen months between the video and its disclosure is not.SecurityThe Fortinet company logo.FortiMail's Exploited Zero-Day Has a Sunday Deadline and No Shipped Patch. Compliance Means Switching Features OffFortinet disclosed CVE-2026-104286, an unauthenticated 9.8-rated flaw in FortiMail already exploited in the wild, and CISA set an October 4 federal deadline the same day. The fixed builds had not shipped, so the only levers before Sunday are turning Identity-Based Encryption off and taking the management interface off the internet.SecurityThe illustration OpenAI published with its report on disrupting a coordinated model-distillation campaign.OpenAI Says Its Hidden Reasoning Was Extracted by Asking the Model to Transcribe It. The Encryption Held; the Model Did NotOpenAI's 30 September report attributes a coordinated reasoning-extraction campaign to individuals associated with Moonshot AI. The numbers are small next to Anthropic's September accusations; the method is not — the operators replayed encrypted reasoning and asked a model to transcribe it, which is the one failure a stronger cipher does not fix.Artificial intelligenceThe Operation Killswitch banner from the law-enforcement takedown notice published by Hamburg police and the Hamburg public prosecutor's office, alongside the partner agency logos.Police Seized 110 Terabytes From KillSec. The Alleged Administrator Is 16 and the Way In Was Cloud StorageEurojust and Europol announced on October 1 that a nine-country operation dismantled the KillSec ransomware-as-a-service group, seizing five servers, its leak site and more than 110 terabytes of stolen data. The suspected administrator is 16, and roughly 500 successful attacks involved no novel exploit at all.Security