AI Act: what actually takes effect on August 2, 2026
On Sunday, another phase of the EU's artificial intelligence regulation begins, and more myths have grown around this date than any other in this law. I break down the timeline into facts: what takes effect now, what's been postponed, what the freshly signed legislation changes - and how to check in a minute whether any of this applies to your company.
legal status as of: 31 July 2026 · this is commentary, not legal advice
Most important to clarify
Half the texts circulating online warn that August 2 brings requirements for high-risk systems: recruitment, scoring, education. This is outdated. The EU's Digital Omnibus package postponed them to December 2, 2027 (Annex III systems) and August 2, 2028 (AI in regulated products), because technical standards weren't ready. What takes effect now are transparency obligations, market surveillance, and real penalties - and those affect a much broader group of companies than it seems.
Timeline on axis
Regulation 2024/1689 starts running, but obligations roll out in stages over four years.
Ban on social scoring, subliminal manipulation and mass facial scraping, among others. Plus mandatory AI competence for personnel.
GPAI provider obligations: documentation, copyright policy, training data summaries.
Art. 50: chatbots must identify themselves, deepfakes must be labeled, emotion recognition disclosed. Market surveillance and real penalties kick in.
Recruitment, credit scoring, education - pushed by Digital Omnibus package from 2026 to late 2027.
Medical devices, machinery, toys with AI - final stage of the timeline.
Does the AI Act apply to your company? Three questions
Do customers talk to a chatbot or voicebot at your business?
Do you publish AI-generated content: images of real people, video, informational texts?
Are you building or selling your own AI system (not: using someone else's)?
Polish Act and KRiBSI
On July 24, 2026, the president signed the AI Systems Act, implementing the EU AI Act domestically. The AI Development and Safety Commission is being established: the chair is to be appointed by October, and the commission should be operational in November 2026. The Commission will handle complaints from citizens and companies about AI systems and issue individual opinions - official confirmation that your solution complies with the law. Micro, small and medium enterprises will get free access to regulatory sandboxes. Penalties: up to €35 million or 7% of turnover for prohibited practices, up to €15 million or 3% for high-risk requirements.
Frequently Asked Questions
What actually starts applying on August 2, 2026?
Three things: transparency obligations from Art. 50 (chatbots must disclose they're AI; deepfakes and some AI-generated content must be labeled; emotion recognition and biometric categorization require notification), market surveillance provisions, and penalties, including Commission penalties for general-purpose model providers.
Do high-risk system requirements take effect on August 2?
No - and this is the most repeated mistake in texts about the AI Act. The Digital Omnibus package pushed standalone high-risk systems from Annex III to December 2, 2027, and AI embedded in regulated products to August 2, 2028. Reason: technical standards weren't ready in time, and without them conformity assessment can't be conducted.
What penalties apply for violating the AI Act?
For using prohibited practices: up to €35 million or 7% of global turnover, whichever is higher. For failing to meet requirements for high-risk systems: up to €15 million or 3% of turnover.
What does the Polish AI Systems Act change?
The law signed by the president on July 24, 2026 establishes a national supervisory authority: the Commission for AI Development and Safety (CAIDS). The Commission will become operational in November 2026, will handle complaints about AI systems and issue individual opinions giving companies legal certainty. Micro, small and medium-sized enterprises will get free access to regulatory sandboxes.
I run a small business and use ChatGPT - does the AI Act apply to me?
Simply using ChatGPT or Claude for internal work doesn't impose obligations on you. The line is drawn where AI touches the customer: a chatbot on your website must identify itself as AI, and published deepfakes and certain generated content must be labeled. Take the quiz on this page - in one minute you'll see which obligations apply to you.
Sources: LEX: AI Act from 2 August 2026 · National Chamber of Legal Advisers · Official source: presidential signature on AI Systems Act. This page is journalistic commentary, not legal advice.
// follow
New tests and price changes, straight to your feed reader
Free RSS, no account and no email needed.
Follow Quantslant →