Latest in AI
Every brief, newest first. Each one links to its primary source.
-
Patch diffing explained in 2 minutes
How comparing software before and after a security fix can reveal the flaw, and why AI is making that faster for attackers.
-
AI regulatory sandboxes explained in 2 minutes
What an AI regulatory sandbox is, why the EU just set common rules for them, and what taking part does and does not mean.
-
How Do AI Image and Video Generators Work? Diffusion Models Explained
How AI turns a line of text into a picture or video clip by starting from static and cleaning it up, step by step.
-
What Is AI Safety? Alignment, Reward Hacking and Oversight Explained
How people try to make AI do what we actually want, why that is harder than it sounds, and what this week’s news says…
-

America’s biggest data centres used about 1.8% of US electricity in a year, Harvard-led study finds
A Nature Sustainability study puts 403 US hyperscale data centres at 68 to 99 terawatt-hours a year, more than half of it from fossil…
-

Drones can now be cleared to fly out of sight from the US coast over open sea, starting with a test run to an offshore platform
The FAA has cleared its first drone flight beyond the pilot’s sight from the US coast over open sea: a test mission to a…
-

AI is making software flaws faster to find and exploit, and slow patching is the weak spot, EU cyber agency says
The EU’s cyber agency says AI is cutting the time to find and exploit software flaws to hours, while many IT providers still take…
-

Driverless Waymo rides are off US freeways after California’s regulator cites an ‘imminent hazard’
California’s DMV barred Waymo’s robotaxis from state freeways after repeated construction-zone incursions, and Waymo has paused freeway rides in its other US cities too.
-

ChatGPT’s newest model is the first OpenAI has released that it rates High for AI self-improvement risk
GPT-6.2 Astra and Sol reach paying ChatGPT users and developers with a lower Astra price, tighter limits on hacking help and new risk ratings.
-

Attackers are using AI that turns security patches into working exploits with no human involved, Google says
Google’s threat researchers found an AI tool that reads new security fixes, writes attacks for the flaws and launches them automatically.
