This weekly update brings major news about AI agent rules and laws around the world. Several countries are working hard to create new rules for smart AI systems that can make decisions on their own.

The European Union made big progress this week. A company called Scytale announced on September 4th that it now helps businesses follow the EU AI Act. This law is the world's first complete set of rules for AI systems. The EU AI Act puts AI systems into four groups based on how risky they are. Some AI is banned completely, while other AI just needs to tell people when they're talking to a robot.

Meanwhile, the United Nations created two new groups to help countries work together on AI rules. These groups will help world leaders make smart choices about AI. One group will study the science behind AI, while the other will let countries talk about AI problems together.

China is also making new AI rules. The country announced new safety standards for AI that creates content like text and pictures. These rules will start working in November 2025. China is also spending $15 billion on quantum computers and AI technology.

One big challenge this week involves agentic commerce - when AI agents buy things for people without human help. Right now, there are no clear laws about what happens if an AI agent makes a bad purchase or gets tricked by criminals. Companies and lawyers are worried about who is responsible when AI agents make mistakes.

Experts say businesses can't wait for governments to make all the rules. Companies need to create their own safety systems now to protect themselves and their customers. The AI world is moving very fast, and the law is trying to keep up.