EU AI Act — August 2026: What Still Applies and What Was Deferred
The EU AI Act is the world's first comprehensive AI law, in force since 1 August 2024. It covers AI placed on the EU market, or AI that affects people in the EU. That includes API, SaaS, or downstream integration, no matter where the company is based.
If your AI system has users in the EU, this part is live now.
Article 50 took effect on 2 August 2026, on the original schedule. The Digital Omnibus did not change it.
You must tell people when they are dealing with an AI system rather than a person. You must also mark AI-generated audio, images, video and text as AI-generated.
This applies to the companies that build these systems and to the companies that deploy them. Enforcement powers started the same day.
There is one grace period. If your generative AI system was already on the market before 2 August 2026, you have until 2 December 2026 to meet the marking and detection duty in Article 50(2).
Anything put on the market after 2 August 2026 had to comply straight away.
- 19 Nov 2025 — proposed by the Commission
- 7 May 2026 — provisional political agreement
- 16 Jun 2026 — approved by the European Parliament (423 for, 57 against, 174 abstentions)
- 29 Jun 2026 — adopted by the Council
- 8 Jul 2026 — signed
- 24 Jul 2026 — published in the Official Journal
- 27 Jul 2026 — entered into force
What it changed: standalone high-risk systems under Annex III moved from 2 August 2026 to 2 December 2027.
AI built into regulated products under Annex I moved to 2 August 2028. Article 50 was left alone.
Article 5 prohibitions — new prohibition added in 2026
Some uses of AI are simply banned. Since 2 February 2025, Article 5 has outlawed:
- Social scoring.
- Scraping faces from the internet to build recognition databases.
- Manipulative AI.
Regulation (EU) 2026/1744 added one more ban: AI that generates non-consensual intimate images ("nudifier" apps) and AI-generated child sexual abuse material.
Phased timeline
- 1 Aug 2024DoneAct entered into force
The EU AI Act entered into force EU-wide.
- 2 Feb 2025DoneProhibited AI practices banned (Article 5)
Social scoring, mass biometric surveillance, and manipulative AI prohibited.
- 2 Aug 2025DoneGeneral-Purpose AI model obligations apply
Documentation, copyright compliance, and systemic risk evaluation obligations begin for GPAI models.
- 2 Aug 2026DoneArticle 50 transparency obligations took effect
Providers must tell users they are interacting with an AI system. They must also mark AI-generated audio, images, video and text. Deployer-facing obligations apply from the same date. Not amended by the Digital Omnibus.
- 2 Dec 2026UpcomingWatermarking grace period ends (Article 50(2))
Some generative AI systems were already placed on the EU market before 2 August 2026. They have until 2 December 2026 to meet the Article 50(2) marking duty. For systems placed on the market after 2 August 2026, the obligation applied immediately.
- 2 Dec 2027UpcomingAnnex III high-risk AI system obligations apply
Deferred from 2 August 2026 by Regulation (EU) 2026/1744 (Digital Omnibus on AI): conformity assessment, technical documentation, human oversight, EU database registration.
- 2 Aug 2028UpcomingHigh-risk AI in regulated products (Annex I)
Deferred to 2 August 2028 by Regulation (EU) 2026/1744 (Digital Omnibus on AI).
What this means for Indian companies
- Article 50 started on 2 August 2026. If your AI system has EU users, you must tell them it is AI, and mark what it generates. Both are law now.
- Annex III moved to 2 December 2027. Four things sit on that later track:
- Your conformity assessment.
- Your technical documentation.
- Appointing an EU representative.
- Registering in the AI database.
- Do not treat the delay as a reason to stop. Documentation, risk classification and governance take 6 to 18 months to build properly.
Check your readiness
Serving users in more than one jurisdiction? Run the readiness checklist to see your baseline DPDP coverage before adding foreign-law obligations on top.
Run it now