EU Unveils AI Act Proposal, Bringing Machine Translation into the Regulatory Frame

On April 21, 2021, the European Commission formally proposed the Artificial Intelligence Act — the world's first comprehensive legislative framework for AI. The bill regulates AI applications by risk tier: unacceptable-risk uses are banned, high-risk uses face strict compliance requirements, and limited-risk uses chiefly carry transparency obligations.

Machine translation is not on the "high-risk" list, but the transparency principle the Act establishes matters deeply to the language industry: when content is generated or processed by an AI system, users should be informed. This points directly at commercial applications of MT and AI-assisted writing — providers of machine translation in the EU market may in future need to disclose that content was machine-generated.

From proposal to final adoption (which came only in 2024), the Act went through a long legislative journey, but its signal was clear from day one: the era of unchecked AI growth is ending, and transparency, traceability and human oversight will form the compliance baseline for AI applications.

For the language industry, regulation is better read as stratification than constraint — providers able to document a compliant "human review + machine assistance" chain will command a premium among compliance-sensitive legal, medical and financial clients.