BRITAIN’s rail regulator, the Office of Rail and Road (ORR), has published an action plan setting out how it will enable the safe adoption of artificial intelligence (AI) by the rail sector. The plan contains six actions which are due to be completed by end of 2026-27.

“Our focus is on ensuring that existing safety and economic frameworks are applied in a way that is proportionate, outcomes-focused and compatible with the use of AI, rather than introducing new, technology-specific rules,” ORR says. “Our role is not to regulate AI itself, but to ensure that the use of AI for regulated activities remains transparent, governable, auditable and consistent with regulatory outcomes and statutory obligations.”

The plan contains six actions, the first of which requires ORR to publish a digital safety strategy and strategic risk chapter by the end of the second quarter of 2026-27. Strategic risk chapters indicate where ORR and the rail industry should maintain strategic focus, based on key risks. The new chapter will emphasise the importance of duty holders understanding how digital systems may affect their operations, and ensuring that they have effective controls in place.

Aiming to complete work by end of the third quarter of 2026-27, ORR will update guidance on how legal duties under The Railways and Other Guided Transport Systems (Safety) Regulations 2006 (ROGS) and the common safety method for risk evaluation and assessment apply to the development and deployment of digital technology, including AI.

A post-implementation review of ROGS, seeking to determine if the regulations provide an appropriate regulatory framework to support the transition to greater reliance on digital technology and AI, including robotics and autonomous systems, has identified low awareness of the resources available to support the industry. ORR says the updated guidance will clarify how the existing legal framework applies to the development and use of AI, and will raise awareness of new and emerging standards, case studies and toolkits available to provide support.

ORR will focus on ensuring that its regulatory processes remain compatible with AI-enabled products by gaining a clear understanding of the impact of AI. This is in line with the AI action plan published by industry innovation and technology body GBRX in April, which says that embedding AI into assurance and regulatory frameworks will reduce uncertainty and support safe adoption, without the need for wholly new assurance models.

Interoperability authorisation

By the end of 2026-27, ORR will report on how AI is being used across the interoperability authorisation process under the Railways (Interoperability) Regulations 2011, confirming that new or upgraded rolling stock and infrastructure meets common technical standards. This will include investigating how AI tools could support the quality and pace of ORR’s own assessment of applications, and considering the use of AI by assessment bodies to streamline evaluation.

ORR will also explore how AI could be used to support its passenger-focused monitoring including complaints, passenger assistance and related research datasets, enabling data and other regulatory information to be analysed more quickly and consistently. This action is due for completion by the end of the second quarter of 2026-27.

Working with data suppliers, ORR will consider whether significant anonymised datasets or synthetic datasets could be made available to support the development, deployment and adoption of AI, examining the barriers to making data publicly available. Target completion date is the end of the third quarter of 2026-27.

ORR’s final action, due for completion by end of 2026-27, is to report on how innovative methods could support AI in areas where different approaches to regulation could deliver benefits.

“ORR’s action plan illustrates how we’re playing our part in enabling responsible adoption of AI with users at the heart of innovation,” says Stephanie Tobyn, director of strategy, policy and regulation at ORR