EU’s ‘AI Omnibus’ Takes Effect, Easing Deadlines but Tightening the Net on High‑Risk AI

On July 27, 2026, the European Union’s new “AI Omnibus” regulation entered into force, updating and streamlining parts of the bloc’s landmark AI Act. The measure does not replace the AI Act; instead, it extends several key deadlines and simplifies administrative requirements so regulators and companies have more time to prepare for enforcement, especially around high‑risk AI systems.

Under the AI Act, applications like biometric identification, medical diagnostics and other systems that can significantly affect people’s rights are classified as “high‑risk” and must meet strict safety, governance and transparency rules. The AI Omnibus adjusts timelines for measures such as national “AI sandboxes” – supervised testbeds where companies can trial new systems – to make it easier for member states to stand them up without freezing innovation.

At the same time, the EU is moving ahead with new transparency obligations for certain AI systems, including some generative AI tools, which begin applying from August 2, 2026. Providers will need to disclose AI involvement more clearly and explain associated risks to users. In contrast to the United States’ more industry‑led approach, Brussels is doubling down on rights‑first regulation, positioning the EU as a global reference point for “responsible AI” – and forcing any company that wants to do business in Europe, including Japanese firms, to adapt their products and compliance processes accordingly.