The European Union has recently released the 'General-Purpose AI Code of Practice' to help businesses comply with AI regulations, focusing on transparency, copyright, and the safety of high-risk AI systems. This code is voluntary and serves as a compliance guideline ahead of the formal implementation of the AI Act, with violations subject to fines up to €35 million (approximately NT$1.225 billion).

The code was officially announced on July 10, aiming to assist thousands of companies in preparing for upcoming compliance requirements. It emphasizes three core areas: 1, enhancing supplier transparency by requiring detailed documentation and information; 2, establishing copyright protection in accordance with EU intellectual property standards; 3, implementing safety management for advanced high-risk AI systems.

Henna Virkkunen, Vice President of the European Commission, described this as a significant milestone in promoting both innovation and safety. Furthermore, companies that opt not to sign this code may find themselves burdened with more complex and costly compliance procedures. The AI Act categorizes AI systems for regulation based on risk levels, with some high-risk applications facing outright bans.

Currently, major tech companies like Google and OpenAI are reviewing the guidelines and have not yet decided whether to sign. These regulations specifically target companies developing general-purpose AI systems, including OpenAI, Google, Microsoft, and Meta. The General-Purpose AI Code of Practice emphasizes three aspects: transparency, copyright, and safety.