EU Reaches Political Agreement on the AI Act

After marathon trilogue negotiations, the European Commission, Parliament and Council reached political agreement on the Artificial Intelligence Act late on December 8, 2023. The world's first comprehensive AI law was settled one year after ChatGPT's release — a legislative pace rare in EU history, reflecting the urgency of the AI transformation.

Among the agreed provisions, those most relevant to the language industry include: providers of general-purpose AI models must meet disclosure obligations including technical documentation and summaries of copyright compliance; generative AI content must be labeled in machine-readable form; and high-risk AI applications face strict compliance requirements. The Act will take effect in phases after formal adoption, with fines for violations of up to 7% of global turnover.

For the language services industry, the "transparency principle" the Act establishes will gradually become a de facto global standard: clients have the right to know whether a translation was AI-generated and what human review it underwent. Forward-looking LSPs have already begun writing "AI use disclosure" into service agreements — compliance capability is becoming a new competitive dimension.

From the 2021 proposal to the 2023 agreement, the EU AI Act's legislative journey ran fully parallel to the generative AI explosion. In an era of technology running at full speed, the anchors of rules are dropping into place.