EU AI Act: What changes as new transparency rules take effect from August 2

The European Union's AI transparency rules become applicable from August 2, bringing mandatory disclosures for certain AI-generated content and significant penalties for violations

  From August 2, new transparency rules for AI-generated and AI-manipulated content will come into force across the European Union. Under Article 50 of the Artificial Intelligence Act, certain AI-generated content, including deepfakes and AI-generated text on matters of public interest, will have to be clearly labelled.

 

The European Commission has also published a voluntary Code of Practice on transparency of AI-generated content to help providers and deployers meet these obligations.

 

As the AI Act applies to providers and deployers whose systems or services are placed on the EU market or used in the EU, the impact is likely to extend well beyond Europe. Meta and Google have signed the voluntary code, signalling their intent to align with the framework.

 

What is the EU AI Act?

 

The EU AI Act is the world’s first comprehensive AI law, adopted in 2024. It uses a risk-based model: the higher the risk an AI system poses to health, safety, or fundamental rights, the stricter the legal obligations. Under that framework:

 

• Unacceptable-risk AI systems are banned
• High-risk systems face strict compliance, documentation and oversight rules
• Certain AI systems, including many generative tools, are subject to transparency obligations
• Minimal-risk systems face few or no specific requirements

 

Article 50’s transparency rules are part of the Act’s phased rollout and are designed to help people recognise when they are interacting with AI or viewing AI-generated content.

 

What must be labelled under the new rules

 

The European Commission groups the main transparency duties into three categories:

 

• Deepfakes
• AI-generated public-interest text
• Disclosure that a user is interacting with an AI system such as a chatbot

 

For compliance with Article 50, the disclosure must be clear, prominent, and easy for users to recognise. The Commission’s voluntary code recommends consistent labels such as “AI”, “AI-Generated” and “AI-Modified” to improve uniformity across platforms.

 

Exceptions and penalties

 

There are limited exceptions. Deepfakes used in clearly artistic, satirical, fictional or similarly creative works can use a lighter disclosure approach, so long as the notice does not spoil the experience.

 

AI-generated text is exempt if it has undergone human review or editorial control and a legally responsible person or organisation accepts editorial responsibility.

 

The AI Act also exempts certain disclosures where AI is used by law enforcement or other authorities to detect, prevent, investigate or prosecute criminal offences.

 

Non-compliance with Article 50 can lead to administrative fines of up to €15 million (about $17.08 million) or 3 per cent of global annual turnover of the AI provider or deployer, whichever is higher.

 

How do India's rules compare?

 

India does not yet have a dedicated AI law comparable to the EU AI Act. Instead, it regulates AI-generated content primarily through amendments to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021.

 

Earlier this year, the Ministry of Electronics and Information Technology amended the IT Rules to require users and social media platforms to prominently label synthetically generated information.

 

However, the two approaches differ in their focus.

 

The EU framework places responsibility on AI developers and providers by requiring transparency to be built into AI systems themselves. It also defines specific exemptions for artistic works and content that has undergone human editorial review.

 

India's approach, on the other hand, places greater emphasis on social media platforms and users. Users are expected to disclose AI-generated content, while platforms are responsible for enforcing the rules and removing content that violates the IT framework.