AI transparency and labelling rules: the changes businesses need to know

The EU’s new AI transparency and labelling rules take effect on 2 August 2026. From that date, businesses launching artificial intelligence (AI) systems must comply with new transparency obligations. The rules also apply to deployers of AI-generated or AI-manipulated content. Businesses already providing AI systems have until 2 December 2026 to comply.

These requirements arise under Article 50 of the EU AI Act. They apply to businesses operating in, targeting, or whose AI outputs are used within the EU. As a result, UK businesses providing AI systems or AI-generated content to EU customers or users will also need to comply. Global businesses should map their distribution channels and downstream uses to identify where EU-facing disclosures and labelling are required.

The rules

Article 50 introduces transparency obligations designed to protect users and the public. Businesses must disclose when individuals are interacting with an AI system, ensure AI-generated or AI-manipulated content is appropriately labelled, and implement measures to identify and signal synthetic content.

The objective is to prevent deception, promote accountability, and enable people to make informed decisions about AI-generated experiences and media.

Compliance

The European Commission has published Guidelines:

• Transparency obligations for providers and deployers of certain AI systems;
• Transparency obligations for providers and deployers of AI systems; and
• EU Icons for labelling AI-generated content.

It also published a voluntary Code of Practice to support compliance with Article 50. Although voluntary, adherence to the Code of Practice provides a recognised framework for consistent labelling and watermarking and may help demonstrate compliance with Article 50.

Consequences of non-compliance

Failure to comply with the EU AI Act may result in regulatory enforcement, financial penalties and remediation orders. Businesses should also consider contractual consequences, as many commercial agreements include warranties, indemnities, audit rights and termination provisions relating to legal compliance.

Non-compliance may also increase exposure under consumer protection, advertising, data protection and discrimination laws. In addition, reputational damage can be immediate where undisclosed AI-generated or synthetic content misleads or undermines customer trust.

Practical steps

Compliance will require a combination of governance, contractual measures and technical implementation. Businesses should:

• Understand the scope of Article 50 and review the European Commission’s Guidelines and Code of Practice.
• Identify where obligations apply across the business, whether as an AI provider or deployer, and develop a documented compliance plan covering in-scope AI use cases, content types and downstream EU distribution.
• Train relevant employees, contractors and agencies on the new requirements.
• Review and update customer terms and conditions, implement appropriate labelling throughout the content lifecycle, and use the EU AI icons where applicable.
• Where providing AI systems capable of generating synthetic content or data, update contractual documentation and provide clear guidance to users on transparency obligations.
• Implement technical measures such as metadata, visible labels and watermarking, together with resilience measures to maintain their effectiveness.
• Update contracts to allocate transparency responsibilities and include appropriate warranties, indemnities, audit rights and termination provisions.
• Monitor regulatory developments, establish incident response and remediation procedures, and consider adopting the voluntary Code of Practice to demonstrate a proactive approach to compliance.

If you have questions or concerns about the new AI transparency and labelling rules, please contact Rebecca Steer.