Society and work

What applies under the EU AI Act from August 2026?

· By Robert Scales · 4 min read

Short answer

August 2, 2026 remains the general application date of the EU AI Act, so the transparency duties for chatbots, deepfakes and AI-generated content start then. The high-risk rules that were due the same day have been postponed by the Digital Omnibus on AI, published on July 24, 2026: to December 2, 2027 for stand-alone systems and August 2, 2028 for AI built into regulated products.

The EU AI Act, formally Regulation (EU) 2024/1689, entered into force in August 2024, but its rules switch on in stages. Article 113 sets the general date of application at August 2, 2026, which is this coming Sunday. For months, that date was expected to bring the bulk of the obligations, including the detailed requirements for high-risk AI. That changed in July.

On June 29, 2026, the Council of the EU gave its final approval to a package called the Digital Omnibus on AI, part of the EU's effort to simplify its rules. It was signed on July 8 as Regulation (EU) 2026/1744, published in the Official Journal on July 24 and entered into force three days later. Its main effect is to postpone the high-risk rules while leaving the general date in place. This article explains what the official texts now say; it is general information, not legal advice.

Two roles to know

The Act assigns duties by role. A provider develops an AI system or general-purpose model, or has one developed, and offers it under its own name. A deployer uses an AI system in a professional capacity; a private person using a chatbot at home is not a deployer. The Act reaches providers outside the EU that sell into the EU, and can also apply when the output of an AI system is used in the EU.

What starts on August 2, 2026

The main obligations that take effect are the transparency duties in Article 50:

DutyWhoWhat it means in practice
Tell people they are dealing with AIProviders of systems that interact with peopleA chatbot must make clear it is a machine, unless that is obvious from context
Mark AI-generated contentProviders of systems that generate audio, images, video or textOutputs must carry a machine-readable marker so they can be detected as AI-made
Inform people exposed to emotion recognition or biometric categorisationDeployersPeople must be told the system is operating
Disclose deepfakesDeployersRealistic AI-generated or altered images, audio or video must be labelled; artistic or satirical works need a lighter disclosure
Disclose AI-written public-interest textDeployersText published to inform the public must be flagged as AI-generated, unless a person reviewed it and someone holds editorial responsibility

The information must be given clearly, at the latest at the first interaction or exposure. One adjustment from the Omnibus: generative systems already on the market before August 2, 2026 have until December 2, 2026 to add the content marking.

The general date also brings the European Commission's power to fine providers of general-purpose AI models under Article 101, which had been held back when the rest of the model rules began in August 2025.

What has moved

RuleOriginal dateNew date
High-risk systems listed in Annex III (such as uses in hiring, education and access to essential services)August 2, 2026December 2, 2027
High-risk AI in products covered by EU product law (Annex I)August 2, 2027August 2, 2028
National AI regulatory sandboxes operationalAugust 2, 2026August 2, 2027
New ban on AI that generates non-consensual intimate images or child sexual abuse materialNewDecember 2, 2026

The Omnibus explains the delay by the late arrival of technical standards, guidance and national authorities, which made the original date hard to meet. Earlier rules are unchanged: the banned practices and the AI literacy duty have applied since February 2025, and the obligations for general-purpose AI models since August 2025. The Omnibus did rewrite the AI literacy article. Providers and deployers must take measures to support the AI skills of staff who operate or use AI systems, without having to guarantee a set level, and the Commission is to publish practical examples.

Try it: List every AI tool your organization uses, one per line, and next to each write "provider" or "deployer" and whether it talks to the public, generates images or text you publish, or reads people's emotions. Those three columns map directly onto Article 50.

What should a small business check?

Start with an inventory. A typical small business is a deployer, using tools made by others: a chatbot on its website, an image generator for social posts, an AI writing assistant, perhaps AI features in its hiring software. A few also build tools and so act as providers.

Then work through this short checklist, adapted from the duties above:

  1. Website or messaging chatbot: does it make clear to customers that they are talking to AI? If you built it yourself, that duty is yours as provider; if you bought it, check that the vendor's version does it.
  2. Images, video or audio you publish: are realistic AI-generated or altered pieces labelled as such?
  3. AI-written articles or notices for the public: either label them, or make sure a named person reviews and takes editorial responsibility.
  4. Emotion or biometric tools: if you use any, are people told? Some uses in workplaces and schools are banned outright.
  5. Staff skills: can you show what you do to help staff use AI tools competently, such as a short training session and written guidance?
  6. High-risk uses: if you use AI to screen job applicants or decide on access to services, note it now. Those rules come in December 2027, and the time is useful for asking vendors what documentation they will provide.

The Omnibus also extends some simplifications, such as proportionate requirements and capped fines, to "small mid-cap" companies, which are slightly larger than SMEs. National authorities may offer guidance to smaller firms, and for anything uncertain, especially a possible high-risk use, a lawyer familiar with the Act is the right next step.

Sources

  1. Regulation (EU) 2024/1689 (Artificial Intelligence Act) — Official Journal of the European Union (EUR-Lex), July 12, 2024
  2. Regulation (EU) 2026/1744 (Digital Omnibus on AI) — Official Journal of the European Union (EUR-Lex), July 24, 2026
  3. Artificial Intelligence: Council gives final green light to simplify and streamline rules — Council of the European Union, June 29, 2026

Get the free Guide to Artificial Intelligence

Sign up for the weekly newsletter and receive the free guide: what changed in AI this week, one lesson, one prompt to try. Unsubscribe in one click.

How we handle your address

Back to top

Display

Your choices are saved in this browser only. No data is sent.

Text size
100%
Colours
Reading
More settings
Accessibility statement