EU AI Act 2026: what changes for AI companies this year, and which rules were delayed

For companies offering chatbots and AI content generators in the European Union, the biggest change this year is about one thing: making sure people know when they are dealing with AI.

Since 2 August, transparency requirements under the EU's AI Act have applied to certain AI systems. The European Commission's AI Office and national authorities have also gained powers to enforce parts of the law, according to the Commission.

Which rules apply depends on the product and on the company's role. A business building an AI model has different duties from an organisation that uses an AI tool to publish content.

What must companies tell users?

Under Article 50 of the law, providers must make sure people interacting directly with an AI system know that it is AI, unless that is already obvious.

Providers of systems that generate text, images, audio or video must also mark their outputs in a machine-readable way, so they can be detected as artificial. This technical marking is separate from a visible label for readers or viewers. Some standard editing tools are exempt.

Organisations that use AI to create or alter deepfakes have their own duty to disclose it. AI-generated text published to inform the public on matters of public interest must also be disclosed, unless a person has reviewed it and someone holds editorial responsibility for it.

For example, a company that builds an image generator and a business that publishes a deepfake made with it fall under different parts of these rules.

What about general-purpose AI models?

General-purpose models, which can be used for a wide range of tasks, come with requirements on documentation, copyright and public summaries of the content they were trained on.

Those rules began applying to new models in August 2025. What changed in August 2026 is that the Commission can now enforce them, including with fines. Models already on the market before 2 August 2025 have until 2 August 2027 to comply, according to Commission guidance.

Which rules have been delayed?

The EU's AI Omnibus amendments entered into force on 27 July 2026 and pushed back the timetable for high-risk AI.

Rules for high-risk uses listed in Annex III of the law, which include some systems used in hiring, education and biometrics, now apply from 2 December 2027. Rules for high-risk AI built into regulated products apply from 2 August 2028, the Commission says.

The changes followed business complaints about overlapping rules and paperwork, Reuters reported. Consumer groups criticised the weaker protections, while the tech industry pushed for further changes.

The Omnibus also added a new ban on AI systems that generate non-consensual sexually explicit or intimate content, or child sexual abuse material.

What happens next?

The next deadline is 2 December 2026. Providers of AI content generators placed on the market before 2 August have until then to meet the machine-readable marking rules in Article 50(2). This does not delay every transparency rule.

The new ban on AI-generated non-consensual intimate content and child sexual abuse material also applies from that date, according to the EU's AI Act Service Desk.