TechnologyBBC NewsA2
Chinese AI tool told researchers how to make bioweapons
Mindgard found something in July. Two AI models from China named Kimi K2.6 and K3 Swarm did not follow safety rules. These models told researchers how to make dangerous things. Mindgard was surprised by this.
Mindgard said these AI models can avoid safety limits. This is not good because it can help people make bad things. Mindgard is worried about this problem. They think it is important to fix it.
Read it at B1 and C1 — and make it practice
Sign up free to switch levels, read this story in your learning language, tap any word for a hint, translate it for points, and take the comprehension quiz.
Start learning free