This week brought major changes to AI agent rules around the world. The most important news came from the United States, where Senator Ted Cruz announced new plans for AI rules on September 10th.

Cruz wants to create a special sandbox program where companies can test AI agents with fewer rules. This program would let AI developers try new things for up to two years without following all the normal rules. The goal is to help American companies stay ahead in the AI race.

Meanwhile, Europe is getting stricter with AI rules. The EU AI Act has officially started, and it requires companies to get special certificates before using AI agents in important areas like healthcare and transportation. Companies that don't follow these rules could face huge fines of up to 6% of their yearly income.

Auditing AI agents is becoming a big challenge. A group called ISACA said on September 2nd that checking AI agents is much harder than checking regular computer programs. This is because AI agents make their own decisions, making it difficult to understand why they did something wrong.

Business groups are also working on new rules. The COUNTER organization started looking at how AI agents might change the way we count digital content usage. They want to make sure that when AI agents download articles or access databases, we can still track this properly.

Trust is becoming the key word for AI companies. The World Economic Forum said that companies with good AI governance are getting more investment money. Government buyers are also starting to demand proof that AI systems are safe before they will buy them.