The EU AI Act officially entered its main enforcement phase on August 2, marking the transition of the law from a legislative concept to actual implementation. Brussels is working to determine how to translate the broad AI rules into specific regulations that companies can follow, and it aims to establish itself as a global standard for AI. Although some legal experts criticize the act for its “chaos,” companies like Anthropic have announced that they will implement compliance measures, such as having the Claude model generate watermarked text in future to meet transparency requirements. In addition to the already effective transparency provisions, the details of subsequent implementation of the act are still being finalized. While the EU may not necessarily force global adoption of its laws, this is not a necessary condition for implementing the act.
The EU AI Act will reach its main enforcement deadline on August 2, marking the transition of the law from concept to practical implementation. Brussels will see if it can translate broad AI regulations into rules that businesses can follow and establish global standards for AI. Although legal expert Patrick Van Iek called it “confusing legislation,” the act is expected to become a global standard, even if it doesn’t serve as a direct template for companies to follow. For example, Anthropic has announced that future Claude models will generate watermarked text to meet the requirements of the act. In addition to the transparency provisions that take effect in August, the subsequent implementation of the act is still underway. The EU may not be able to fully adopt its laws globally, but it’s not necessary.