AI View: May 2025

Welcome to AI View, Simmons & Simmons' fortnightly round-up of key AI legislative, regulatory, and policy updates from around the world.

This edition brings you:

1. UK Financial Conduct Authority to launch live AI testing service

2. UK government rejects EU request for EU AI Act to apply fully in Northern Ireland

3. China announces nationwide campaign against AI misuse

4. BRICS Foreign Ministers sign declaration on AI governance

5. Japan's House of Representatives passes the Bill on the Promotion of Research, Development and Application of AI-Related Technologies

6. Colorado fails to pass bill amending Colorado AI Act  

7. Indonesia issues comprehensive framework for banks on AI adoption

1. UK Financial Conduct Authority to launch live AI testing service

On 29 April 2025, the UK Financial Conduct Authority (FCA) announced the upcoming launch of a live AI testing service, which is intended to assist firms in the development and deployment of AI.

The live testing service will provide regulatory support to firms who are ready to deploy consumer or market-facing AI models. Firms will be able to collaborate with the FCA before deploying consumer or market-facing AI tools to ensure that their models are compliant with relevant regulations.

This testing service has a proposed launch date in September 2025, with plans to run for a period of 12-18 months. The FCA is currently seeking views from firms regarding the introduction of this service, with the call for input to close on 10 June 2025.

Read the announcement here.

2. UK government rejects EU request for EU AI Act to apply fully in Northern Ireland

On 30 April 2025, the UK government rejected a request from the European Commission for the EU AI Act to take full effect in Northern Ireland under the Brexit Withdrawal Agreement.

A number of articles within the EU AI Act, such as Articles 103 to 107 and Article 109, have already taken effect in Northern Ireland. The EU AI Act is being implemented in stages and is due to fully apply in the EU in August 2026.

However, the UK government refused to agree to an EU decision adopted in March 2025 that the Windsor Framework (an agreement between the UK and the EU that amends the Northern Ireland Protocol) should be amended to apply the entirety of the EU AI Act in Northern Ireland.

In the statement, the UK government stated that the EU AI Act is "complex and requires further scrutiny" and they have "asked for further dialogue with the EU under the terms of the Withdrawal Act".

Read the UK's statement here.

3. China announces nationwide campaign against AI misuse

On 30 April 2025, the Cyberspace Administration of China (CAC) announced a three-month nationwide campaign to combat AI misuse, including using information that infringes on others' intellectual property rights, privacy rights and other rights.

The campaign will be carried out in two phases. The first phase will focus primarily on removing illegal AI applications, strengthening the regulation of AI-generated content (particularly regarding the use of deepfake programmes, voice synthesisers and face-swapping tools), and encouraging online platforms to improve their ability to identify and verify violations. The CAC has also noted that they will be investigating failures to label AI-powered content.

The second phase will target specific issues such as the use of AI to create and disseminate rumours, misinformation, obscene content, impersonation, and instances of "internet trolling". The CAC stated that the second phase will also focus on removing illegal and harmful content, while penalising violations committed by relevant accounts, multi-channel network organisations and online platforms.

Read more here.

4. BRICS Foreign Ministers sign declaration on AI governance

On 29 April 2025, following a meeting in Rio de Janeiro, the foreign ministers from the BRICS+ countries (Brazil, Russia, India, China, South Africa; now also including Indonesia, Saudi Arabia, Egypt, Ethiopia, Iran, and the United Arab Emirates) issued a declaration on their commitments on AI.

The declaration emphasised BRICS countries' focus on establishing inclusive and international governance for AI and noted that AI systems should be developed and deployed in compliance with national regulatory frameworks and international principles during development and deployment.

The countries agreed that the global governance of AI must mitigate potential risks and address the needs of all countries, especially those of the Global South. The declaration also emphasised the need to ensure that transparency, ethical standards, data protection and equitable development outcomes are championed by AI governance.

Read the declaration here (in Portuguese only).

5. Japan's House of Representatives passes the Bill on the Promotion of Research, Development and Application of AI-Related Technologies

On 24 April 2025, Japan's House of Representatives passed the Bill on the Promotion of Research and Development and Application of Artificial Intelligence-Related Technologies (the Bill), Japan's first comprehensive law on AI.

The Bill mandates state and local governments to advance AI-related technologies by establishing policies and guidelines. The Bill establishes a national framework for AI development, defining AI-related technologies as those replicating human cognitive abilities through artificial methods and information processing systems.

Private sector entities using AI-related technology are required to cooperate with government-led initiatives on AI, but are not subject to specific hard obligations. This obligation is expected to apply to entities that develop, provide, or use AI-based products or services in connection with their business activities. 

The Bill introduces two mechanisms to guide implementation:

• AI Basic Plan: The government will establish fundamental policies for measures to be implemented to promote research, development and utilisation.

• AI Strategy Headquarters: The Headquarters will oversee the AI Basic Plan and direct AI policy across the Japanese government.

Further provisions of the Bill cover issues including research promotion, responsible use of AI, talent development, AI education, monitoring of AI trends, and issue analysis.

The Bill does not provide penalties for non-compliance.

Read more here (only available in Japanese).

6. Colorado fails to pass bill amending Colorado AI Act

On 28 April 2025, Senator Robert Rodriguez introduced a bill (the AI Bill) proposing an amended version of the Colorado AI Act (the Act) to the Colorado General Assembly. Last year, Colorado became the first of the US states to pass comprehensive legislation to regulate AI. Following concerns that the Act might be too far-reaching, the AI Bill was introduced with an aim to narrow the Act's focus and materially reduce the compliance burden on in-scope businesses. However, the AI Bill failed to pass the before the 7 April deadline when the Colorado legislature adjourned.

This AI Bill proposed key changes to the Act, such as:

• Revised definitions and exemptions: Amendments to key definitions, such as "developer", "algorithmic discrimination" and "high-risk AI system".

• Wider exemptions: New exemptions for deployers using high-risk AI systems solely for recruitment, sourcing, or hiring of external candidates.

• Modified obligations: Many of the deployer obligations apply only to high-risk AI systems that make consequential decisions with no meaningful human involvement.

• Delayed enforcement: The Act will not be enforced until 1 January 2027.

The failure of the AI Bill to pass the Colorado House of Representatives in time means that the current version of the Colorado AI Act will continue to enter into force on 1 February 2026 as enacted last year.

The AI Bill can be found here.

7. Indonesia issues comprehensive framework for banks on AI adoption

On 29 April 2025, Indonesia's Financial Services Authority (the OJK) introduced a comprehensive framework that seeks to guide banks to adopt and develop AI responsibly.

This framework outlines the base principles and best practices for banks to follow throughout all stages of the AI lifecycle (from development and testing to implementation and oversight). It also seeks to promote the use of predictive, generative and agentic AI across the banking sector to boost productivity and innovation, notably in relation to credit scoring, fraud detection, product development, and customer service, pricing and market intelligence. However, banks are also warned to be cautious of unchecked usage of AI, and encouraged to maintain audit trails, classify risks and ensure human oversight over AI systems.

The framework is not legally binding. Instead, it will serve as the best practices guidance for banks in Indonesia. Notably, the framework contains the requirement that all AI systems used in banking are aligned with Indonesian national values (most specifically Pancasila, the foundational and official state philosophy).

The guidance can be found here (in Indonesian only).