This week brought major changes to how countries plan to control AI agents - smart computer programs that can make decisions on their own.

In California, Governor Gavin Newsom said he will sign a new law called the "No Robo Bosses" Act by September 30. This law stops companies from using AI systems alone to fire workers or discipline them. Companies must also tell workers at least 30 days before using AI to make choices about hiring, work schedules, or job performance.

The United Nations started a big new project on September 25 to create rules for AI around the world. All 193 countries that belong to the UN agreed to make two new groups: one with scientists to study AI safety, and another to talk about AI rules. UN Secretary-General António Guterres said these groups will help make sure AI technology helps people instead of hurting them.

Experts are especially worried about "agentic AI" - these are AI systems that can set their own goals and make choices without people watching them closely. About 69% of AI experts think these systems need completely new ways to manage them because they're so different from older AI.

The European Union already has strict AI rules that started this year. Companies that break these rules can be fined up to 35 million euros or 7% of all the money they make worldwide. These huge fines are making companies around the world pay more attention to AI safety.

In the United Kingdom, officials published new guidance in September about how different government agencies will watch over AI. Unlike other places, the UK decided not to make one big new law. Instead, they want existing agencies to use five main rules: safety, security, fairness, transparency, and the right for people to challenge AI decisions.