Commission prepares fallback guidelines amid delays to technical standards

The European Commission is drafting contingency guidelines to support implementation of the AI Act, amid concerns that key technical standards may not be delivered on time. The move follows repeated delays in the development of harmonised standards by European standardisation bodies.

Under the AI Act, detailed technical standards are meant to clarify how providers of high-risk AI systems can demonstrate compliance with legal requirements. These standards are being developed by CEN-CENELEC, but several are now expected only by April 2027, close to or beyond planned enforcement milestones.

The delays have triggered calls from some governments to freeze parts of the AI Act until standards are available, arguing that companies cannot comply with high-risk obligations without clear technical benchmarks. In response, the Commission proposed in November to push back the application of high-risk AI rules to December 2027 or August 2028.

According to documents seen by Euractiv, the Commission now plans to issue non-binding guidelines as a transitional measure if standards remain unavailable. These guidelines would aim to clarify expectations for compliance and reduce pressure for further postponements, while stopping short of creating new legal obligations.

The approach differs from “common specifications”, a formal fallback mechanism already included in the AI Act, which would allow the Commission to adopt binding technical requirements if industry-led standards fail to materialise. Instead, the guidelines are intended as a softer, interim solution while standardisation work continues.