On 12 January 2026, the International Regulatory Strategy Group (IRSG) published a report mapping how jurisdictions are approaching artificial intelligence (AI) in financial services. The report identifies where global coherence is emerging and outlining practical steps for policymakers and regulators to promote the safe and responsible innovation of AI.
Key findings
The report draws numerous conclusions regarding the implementation of AI in financial services, including:
- Positive alignment on the high-level principles that should govern AI, with most frameworks drawing on the OECD and G20/G7 endorsed principles of human-centricity, transparency and explainability, robustness and safety, and accountability. Despite this, there is significant global divergence in how these principles are implemented within national approaches, such as the outcomes-focused supervision of the UK, compared to the prescriptive rulebooks adopted by the EU.
- AI is a general-purpose technology which potentially could magnify existing financial sector risks, under certain circumstances, rather than introducing wholly new risks. These risks include model risk, data governance, and third-party concentration. To help mitigate these risks, the IRSG encourages firms to engage in collaboration with authorities to explore good practice within technology-neutral rules, rather than new regulations or guidance that risks rapidly becoming outdated.
- The IRSG sets out recommendations in the report, on how to achieve coordinated, interoperable, and principle-based supervision anchored in existing regulatory frameworks. These should be reinforced by the national authorities and international standard setters.
- The most effective near-term path is to leverage and align existing frameworks, rather than creating AI-specific global rulebooks. The IRSG encourages coherence through shared principles and interoperable supervision, whilst the main drivers of fragmentation, including competition and data localization, should be managed through collaborative, principle-based solutions.