AI Act: 2 August 2026

No avalanche of high-risk requirements: the real deadline was about transparency and enforcement, and it affected far more companies than you might think.

For months, 2 August 2026 was portrayed as the day the AI Act would “explode” into force with all its most demanding requirements. That’s not what happened, and understanding why matters if businesses want to set the right priorities.

What was postponed. Regulation (EU) 2026/1744, published on 24 July and entered into force on 27 July 2026, postponed the application of requirements for high-risk systems listed in Annex III until 2 December 2027. For high-risk systems embedded in regulated products, the new date is 2 August 2028.

What was NOT postponed and affected almost everyone.
→ Transparency obligations (Article 50): companies using chatbots or virtual assistants, or generating synthetic content (text, images, audio), must ensure that this is clearly identifiable.
→ From 2 August, the enforcement regime also extended to obligations that became applicable on that date.
→ No postponement for Public Administrations.

The penalties remain those established under Article 99: up to €35 million (or 7% of global annual turnover) for prohibited practices, and €15 million (or 3%) for breaches of other obligations.

The right question is no longer “When will we have to deal with this?” but “What can we demonstrate today?”.

An inspection, a report, an internal dispute: since August, organisations may be asked to provide concrete evidence of due diligence, rather than statements of intent. An inventory of AI systems currently in use, risk classification, staff training, an obligation already applicable since February 2025, are all part of a governance framework that cannot wait for the official compliance calendar.

Sources: Regulation (EU) 2024/1689, Articles 50 and 99; Digital Omnibus, Council of the EU (29/06/2026)

#AIAct #ArtificialIntelligence #Compliance #GRC #CyberSecurity #DigitalOmnibus #AIGovernance #MagisPartners