This week saw major developments in AI ethics and safety worldwide.

The European Union updated rules for AI systems to make companies more responsible. Meanwhile, the UK made it illegal to create AI-generated child abuse images, with strict new laws.

In security news, Microsoft launched AI cybersecurity agents that can automatically stop hacking attempts. The UK also renamed its AI safety group to focus more on stopping AI-related crimes.

A big meeting in Geneva discussed how military AI should be used safely, with experts talking about accountability and testing.

In the US, some states banned a Chinese AI tool called DeepSeek over spying fears. Doctors faced questions about using AI for patient messages without telling people.

Experts warned that AI agents becoming too independent could cause social problems if people disagree about their "feelings".