AI View - September 2026

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

AI x Dispute Resolution Webinar Series Returns This September

Our AI x Dispute Resolution webinar series resumes after the summer break, with upcoming sessions on AI and IP (3 September) and AI Incident Response & Product Liability (1 October).

The series has already explored AI disputes, regulatory enforcement, collective actions and negligence, providing practical insights for legal, risk and compliance teams navigating the rapidly evolving AI landscape.

Register for upcoming sessions here.

This edition brings you:

1. UK Solicitors Regulation Authority publishes warning notice on misuse of AI in legal services

2. EU AI Office has reportedly already started exercising its enforcement powers under the AI Act

3. South Korea adopts final draft of AI Ethics Principles

4. Japan adopts code of principles for generative AI developers and providers

5. Hong Kong publishes guidance on protecting personal data privacy in use of agentic AI

6. Singapore opens consultation on the impact of AI on intellectual property regime

7. South Korea amends Personal Information Protection Act to allow limited use of personal data for AI development purposes

1. UK Solicitors Regulation Authority publishes warning notice on misuse of AI in legal services

On 17 August 2026, the UK Solicitors Regulation Authority published a warning notice regarding firms' and individuals' professional obligations when using AI in the provision of legal services.

The notice identifies two particular areas of concern:

1. the risk of false or incorrect information in court or other documents as a result of use of AI, especially arising from "hallucinations"; and
2. the risks to client confidentiality, in relation to personal data and other confidential information, when using AI systems.

The notice emphasises that AI has no separate legal personality and that regulated individuals using AI in the course of delivering legal services remain accountable for their work and outputs. Individuals must be mindful of their professional obligations to ensure that the service they provide is competent, with effective supervision of work, while firms must have effective governance structures, systems and controls in place to manage risks.

Read the notice here.

2. EU AI Office has reportedly already started exercising its enforcement powers under the AI Act

On 2 August 2026, the AI Office's enforcement powers under the AI Act became operational, including powers to investigate general-purpose AI (GPAI) providers, request information and documentation, conduct model evaluations (including access to models through appropriate technical means and, where necessary, source code), require corrective measures, and impose fines of up to €15 million or 3% of worldwide annual turnover for GPAI-related breaches.

While national market surveillance authorities remain primarily responsible for supervising AI systems, the AI Office has exclusive competence for GPAI models and certain AI systems under Article 75(1), namely systems: (i) based on GPAI models where the model and the system are developed by the same provider or undertaking (subject to specified carve-outs) and (ii) that constitute or that are integrated into a very large online platform or very large online search engine.

The AI Office is already exercising these powers, having reportedly issued over 30 requests for information to GPAI providers.

Read more on enforcement of the AI Act here.

3. South Korea adopts final draft of AI Ethics Principles

On 24 August 2026, South Korea's Ministry of Science and ICT adopted the AI Ethics Principles (the Principles).

The Principles, which amend the National Guidelines for AI Ethics adopted in 2020, provide voluntary guidance for developers, businesses, and individual users on the responsible use of AI. While non-binding, the Principles are intended as a foundation for further sector-specific ethical standards and guidelines.

The guidance centres around three core values – human dignity, the public good of society, and the sustainability of humanity – which are to be realised through seven principles:

1. Human-centredness
2. Protection of privacy
3. Fairness and inclusiveness
4. Accountability
5. Safety
6. Reliability
7. Transparency

No order of priority is assigned to the selected principles. Rather, actors are encouraged to balance and apply the Principles as they consider appropriate in the given context.

The South Korean government will regularly review and supplement the Principles in light of newly arising issues and periodically obtain stakeholder views to identify difficulties with their implementation.

Read the Principles here (Korean only).

4. Japan adopts code of principles for developers and providers

On 25 August 2026, Japan's Cabinet Office adopted a code of principles for developers and providers of generative AI (the Code), meaning companies which build AI technology and those which make that technology available to users.

The non-binding Code aims to promote the advancement of generative AI while ensuring protection of intellectual property rights. The final version of the Code comes following public consultation with stakeholders earlier this year.

The Code follows a "comply-or-explain" model, under which developers and providers should either implement the principles or explain the reasons for their non-implementation. The scope of the Code covers not only Japanese businesses, but also overseas entities providing AI systems in Japan.

Those who choose to comply with the Code would be expected to:

• announce and notify the government of their compliance;
• publish information about model design specifications, training processes, and data; and
• review and update this information annually.

The Code states that developers and providers would not be required to disclose any sensitive information, such as trade secrets or information relating to safety and security.

Read the Code here (Japanese only).

5. Hong Kong publishes guidance on protecting personal data privacy in use of agentic AI

On 25 August 2026, Hong Kong's Office of the Privacy Commissioner for Personal Data published guidance on protecting personal data privacy in the use of agentic AI (the Guidance).

The Guidance supplements the AI Model Personal Data Protection Framework published by the Privacy Commissioner in June 2024. While the Guidance itself is non-binding, it is intended to help actors comply with the requirements of Hong Kong data protection law.

The Guidance identifies particular data privacy risks which arise due to the degree of autonomy exercised by agentic AI, including:

• greater access rights to personal data;
• inaccurate information being disseminated through multi-agent systems; and
• function creep, meaning the use of data for purposes other than the original intended or authorised purpose.

The Guidance sets out recommendations for both organisations and individual users to address these risks and includes a Security Checklist of practical steps to take throughout the various stages of the AI lifecycle e.g., preparation, deployment, and use.

Read the Guidance here.

6. Singapore opens consultation on the impact of AI on intellectual property regime

On 26 August 2026, Singapore's Ministry of Law (MinLaw) and the Intellectual Property Office of Singapore (IPOS) opened a public consultation on the impact of AI on Singapore's intellectual property regime, open until 22 October 2026.

MinLaw and IPOS have previously engaged with stakeholders including rights owners, technology companies, and representatives from creative industries. This consultation seeks broader views from interested parties on how Singapore's IP laws and frameworks can remain fit for purpose in the context of continuing AI developments.

The consultation focuses on three broad areas in relation to copyright, and two in relation to patents.

For copyright, the areas of focus are:

1. how to foster greater certainty and accountability in AI training;
2. copyright risk management in AI deployment and use; and
3. the nature of human creativity in AI-assisted works.

For patents, the areas of focus are:

1. how existing inventorship principles should apply across the spectrum of human-AI interactions in the inventive process; and
2. how the large-scale publication of AI-generated technical disclosures may affect the prior art landscape, meaning the body of existing public information which determines whether a patent should be granted.

Read the consultation paper here.

7. South Korea amends Personal Information Protection Act to allow limited use of personal data for AI development purposes

On 20 August 2026, South Korea's National Assembly passed an amendment to the Personal Information Protection Act to allow use of personal data for the development of AI in certain circumstances.

Until now, use of personal data for AI development has been permitted only through regulatory sandboxes, meaning limited and supervised "trial runs" approved on a case-by-case basis.

This amendment allows the use of lawfully collected personal data for AI development purposes only where pseudonymised or anonymised data is insufficient, and there is a public interest or social need.

The amendment also introduces safety provisions, such requiring prior assessment of risk factors in cases where there is significant impact or risk to the interests of data subjects.

Read the announcement of the amendment here (Korean only).