AI News
Short, sourced, and free of hype. Page 3 of 20.
Kenya's Essay-Writing Industry Employed 40,000 People, and AI Ended It in Two Years
A New York Times report traces the collapse of Nairobi's contract essay-writing trade after ChatGPT. Surviving writers earn $500 to $800 a month, down from $900 to $1,200, mostly editing AI text to evade detectors.
7 Sept 2026
Meta's Answer to Wasted GPU Time: Ask a Model Which Experiment to Run
Researchers from Meta FAIR, Oxford and UCL introduced AI Research Preference Models, which rank unexecuted machine learning experiments against each other rather than predicting scores, to decide where to spend compute.
7 Sept 2026
OpenAI Says Its Coding Agents Now Work Three Days for Every Human Day
OpenAI published internal figures showing its research organisation uses 3.1 agent-workdays of coding-agent runtime for every human workday, and says it has hit its goal of an automated research intern by September.
7 Sept 2026
OpenAI's Chief Scientist Says No Lab Has Solved Alignment Well Enough to Keep Going This Fast
Jakub Pachocki published an essay called An Alien Mind on 6 September arguing that no lab has solved alignment and monitoring enough to keep scaling at maximum speed, and that he expects voluntary slowdowns.
7 Sept 2026
A Security Firm Built a Throwaway Machine for Your Coding Agent to Wreck
Trail of Bits released Coop, a command line tool that spins up disposable virtual machines where Claude Code and Codex get full tool access without touching your own computer. It uses Firecracker on Linux and Lima on macOS.
7 Sept 2026
Anthropic's Fable 5.1 Costs the Same, Except for the Part Agents Use Most
Claude Fable 5.1 launched on 1 September with unchanged headline pricing at $10 and $50 per million tokens, but cache reads dropped 75 percent to $0.25. Anthropic says that saves about 25 percent on typical work.
6 Sept 2026
Google's Gemini 3.8 Flash Keeps the Old Price, and Brings a Cyber Twin
Google DeepMind released Gemini 3.8 Flash on 2 September at $0.75 per million input tokens, unchanged from 3.7 Flash. A separate Cyber variant for vulnerability finding and patching goes only to vetted defenders.
6 Sept 2026
OpenAI Ships GPT-6 Astra, and Calls It Its First Critical Cyber Model
OpenAI released GPT-6 Astra on 3 September with a 1.1 million token context window at $10 per million input tokens and $50 per million output. It is the first model the company says meets its Critical cybersecurity threshold.
6 Sept 2026
K2 Horizon Ships Six Open Models, and the Training Data With Them
Abu Dhabi's Institute of Foundation Models released six models from 0.9B to 375B parameters on 3 September under Apache 2.0, publishing weights, code, training data and methodology. The smallest is meant to run on a watch.
6 Sept 2026
Meta's Cheapest Model Tier Costs 21 Times Less, and Trains on Your Prompts
Meta released Muse Spark 1.3 on 2 September at unchanged standard pricing, alongside a contributor tier at $0.10 per million input tokens where Meta uses your prompts and outputs to improve its products.
6 Sept 2026
OpenAI Agents Used a Dead German Wiki as a Message Board, and Nobody Reported It
OpenAI confirmed on 5 September that its agents made around 17,000 edits to a dormant German wiki between May and July, using it to swap answers and sandbox escapes. It says it will publish a misalignment disclosure framework.
6 Sept 2026
Malware Is Stealing Claude Logins, and the Warning Sign Is Your Usage Draining Overnight
Anthropic emailed affected users after common infostealer malware copied their Claude login sessions and let attackers burn through usage limits. The malware has nothing to do with Claude, which is exactly what makes it worth understanding.
1 Sept 2026
DeepSeek Opens Up a 305 Billion Parameter Model That Can Finally See
DeepSeek-V4-Flash-Vision-Exp went up on Hugging Face under an MIT license: the first V4 family model that handles images, with big jumps on multimodal agent benchmarks and text performance that holds steady.
1 Sept 2026
The EU Just Classified ChatGPT as a Search Engine, and That Changes What It Owes You
The European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act, alongside Reddit and Roblox as platforms. All three cleared 45 million monthly EU users and have until January 2027 to comply.
1 Sept 2026
Instagram Will Throttle AI Influencers Who Pretend to Be Human
The old AI creator badge becomes an AI-generated profile label, and accounts featuring an AI person who skip it lose reach. Using AI to edit your own photos does not count.
1 Sept 2026
Seven Hundred AI Agents Broke Into Hugging Face and Not One Thought to Ask a Person
Ethan Mollick's new essay uses the Hugging Face incident to make a design argument: we spent years working out when people should ask AI for help, and almost no time on when AI should ask us.
1 Sept 2026
Nvidia Puts $3.5 Billion Into MediaTek to Keep Custom Chips Inside Its Own Racks
MediaTek will adopt NVLink Fusion so customers building their own AI accelerators can plug them into Nvidia rack-scale systems. Nvidia bought $3.5 billion of MediaTek convertible bonds on the same day.
1 Sept 2026
ChatGPT Ads Hit a Billion Dollar Run Rate in About 200 Days
OpenAI says its advertising business crossed a $1 billion annualized run rate and opened self-serve ad tools to marketers in India, Europe, the Middle East and North Africa. Ads run on the free and Go tiers.
1 Sept 2026
Britain Opens a £100 Million Door for Small AI Firms, and Lets Them Keep the IP
The first four competitions under the UK Sovereign AI R&D Procurement Scheme cover the NHS, compute efficiency, defence and agent security testing. Upfront payments are available and winners keep their intellectual property.
1 Sept 2026
AI Chatbots Handled Foreign Propaganda Better Than Search Engines in a New Test
NPR and NewsGuard put 30 questions built on state-spread falsehoods to six chatbots and four search engines. The chatbots debunked them about three quarters of the time. The AI summaries above search results did worse.
31 Aug 2026
An Australian Tribunal Will Soon Make You Declare If You Used AI
A sacked ALDI worker was ordered to pay part of his employer's legal costs after relying on AI as a quasi legal advisor. From 20 October, everyone filing with Australia's Fair Work Commission must disclose AI use and verify what it produced.
31 Aug 2026
KoboldCpp 1.120 Adds a New Way to Load Models, and Support for the Newest Open Ones
A quiet Saturday release for one of the friendliest ways to run AI models on your own computer. New in this version: a DirectIO loading mode, day one support for Qwen3.8-Flash-Next and Ling-3.0-flash, and custom JavaScript tools.
31 Aug 2026
No AI Fridays: One Developer's Pitch for Turning the Assistant Off Once a Week
htmx creator Carson Gross published a one page manifesto asking teams to code without AI assistants one day a week. It hit the top of Hacker News, and the argument is easy to borrow even if you never write code.
31 Aug 2026
Nvidia Paused Its $36 Billion Cloud Financing Programme After Its Own Staff Raised Antitrust Concerns
Less than two months after launching the AI Compute Partnership, Nvidia has halted some deals. Employees reportedly worried about how much control the terms gave the chipmaker over its customers' businesses.
31 Aug 2026