Good hacker, bad hacker: Alibaba says it trained its new flagship AI model, Qwen4-Max, so it can help programmers patch security holes, but won’t help hackers find new ones to exploit, Reuters reports. It’s the latest Chinese lab to show its safety work, after Tencent and Moonshot.
Check the specs: South Korea will inspect every pair of glasses at this week’s university entrance exam, Reuters reports. Blame the student who came first in this year’s tax accountant exam, then confessed he’d snapped the questions with a hidden camera and had an AI tool answer them.
Killer robots: Most countries want a treaty that limits AI-powered weapons, like drones that can choose and attack targets without a human giving the order. But the US and Russia look set to block the push at a UN meeting in Geneva this week, Reuters reports. They argue that existing laws of war are sufficient to prevent such actions.
The S-tier of the “machines making mischief” movie genre obviously includes 2001: A Space Odyssey and Dr. Strangelove. Slightly off the beaten track, check out Colossus: The Forbin Project.
That’s it for today! Like what you’re reading? Feel free to invite some friends. Or drop us a line at [email protected]. We’ll be back in your inbox on Thursday.