This weekly update covers major changes in legal rules for AI agents around the world.

The European Union released new guidance on how companies must follow the EU AI Act when using agentic AI systems. These are AI tools that can work on their own without human help. Companies now face tough questions about how to classify their AI agents and what safety rules they must follow.

In the United States, the government took a very different path. The White House published its "America's AI Action Plan" on July 23, showing a move toward fewer regulations. This approach focuses on letting private companies lead AI development with less government oversight. However, individual states are still making their own AI laws, creating a patchwork of different rules across the country.

Antitrust concerns are growing as agentic AI systems become more powerful. Legal experts warn that AI agents could accidentally help companies work together in ways that hurt competition. This happens when AI systems share private business information that leads to unfair practices.

Several major companies launched new legal AI tools this week. Thomson Reuters introduced CoCounsel Legal, which uses agentic AI to help lawyers with research and case work. Meanwhile, corporate law departments are finding new ways to use AI agents for tasks like monitoring regulations and compliance.

The healthcare industry is being extra careful about using AI agents in clinical trials. Regulators are working with companies to figure out which uses are safe and which need more oversight. Simple tasks like scheduling can use AI agents easily, but decisions about patient care need human doctors to stay involved.