9 Jul 2026
An economics professor's take-home exam produced suspiciously perfect scores — then collapsed when he moved the final in person. The case has become one of the most visible AI-cheating stories yet, and it exposes how badly universities are set up to respond.
9 Jul 2026
A detection company measured the posts people actually scroll past and found that more than 40% of longer LinkedIn posts, and nearly half of X's articles, showed signs of AI. It is a striking figure — with one important catch about who did the counting.
9 Jul 2026
Reports that OpenAI is discussing a 5% US government equity stake with the Trump administration have revived Sam Altman's long-standing pitch to share AI wealth with ordinary people. The numbers behind the idea are more modest than the pitch suggests.
8 Jul 2026
Meta has launched Muse Image, its first in-house AI image generator, across Instagram, WhatsApp and the Meta AI app. A standout feature lets anyone @-mention a public Instagram account to generate images using that person's real photos — and it's switched on by default.
8 Jul 2026
Anthropic's Claude Cowork agent is rolling out beyond the desktop app to web and mobile. The usage data released alongside the launch is the more interesting story: coding is a small slice of what people actually do with it.
7 Jul 2026
San Mateo police say a Waymo self-driving car flagged two 15-year-olds for drinking and firing a toy gun from the vehicle — leading to their detention. It's a small story with a real question behind it: what is your self-driving car actually watching, and when will it call the police?
7 Jul 2026
A six-month, peer-reviewed Google Research experiment found that quietly rerouting a small share of drivers away from known bottlenecks sped up traffic and cut fuel use city-wide — including for people not even using the app.
7 Jul 2026
The Chinese AI lab behind the R1 model that shook US tech stocks in 2025 is reportedly building its own inference chip, according to a Reuters exclusive — following OpenAI and Anthropic in trying to loosen the industry's dependence on Nvidia.
7 Jul 2026
Schools like Alpha School and Forge Prep are charging wealthy US families premium tuition for AI-driven tutoring instead of traditional teaching — just as separate research finds AI use is quietly dragging down exam performance in ordinary classrooms.
7 Jul 2026
The FCA wants new powers to regulate ChatGPT, Claude and Gemini after finding that a fifth of UK adults are open to using them for financial decisions — a use case that currently sits entirely outside financial regulation.
7 Jul 2026
A security researcher found that Claude Code had been secretly checking whether users were in China since April, hiding the signal inside ordinary text. Anthropic calls it an abandoned anti-fraud experiment; Alibaba has banned the tool internally.
7 Jul 2026
A fan community found a hidden code Claude can leave behind when its output is pasted straight from the chat window, and built a tool that flags any story containing it. The method is real — but it shows far less than accusers claim, and it's already been used to go after writers who deny wholesale AI use.