01 / Signal
From 2 August 2026, Article 50 applies to relevant providers and deployers, subject to its scope and exceptions, and the Commission can enforce applicable general-purpose AI-model obligations, including through fines. Regulation (EU) 2026/1744, in force since 27 July 2026, moved the Chapter III high-risk rules to 2 December 2027 for Article 6(2) and Annex III systems and to 2 August 2028 for Article 6(1) and Annex I product-related systems.
02 / The timetable at a glance
- 2 August 2025
- Governance rules and obligations for providers of general-purpose AI models became applicable.
- 2 August 2026
- Article 50 transparency obligations apply, and the Commission can enforce applicable general-purpose AI-model obligations.
- 2 December 2027
- The revised application date for Article 6(2) and Annex III high-risk AI systems.
- 2 August 2028
- The revised application date for Article 6(1) and Annex I product-related high-risk AI systems.
03 / A layered timetable, not a general postponement
The AI Act did not begin on a single date. Prohibitions, definitions and AI-literacy provisions have applied since February 2025. Governance rules and the obligations for providers of general-purpose AI models followed in August 2025. The 2 August 2026 milestone adds Article 50 transparency duties and the Commission’s ability to enforce applicable general-purpose AI-model obligations. Leadership therefore needs a position that distinguishes what already applies from what has moved.
That distinction matters commercially as well as legally. A board, investor, customer or counterparty may ask about an organisation’s AI role, public claims or model dependencies before the later high-risk dates arrive. A postponed category date is not evidence that every present exposure is postponed.
04 / Article 50 reaches public interactions and generated content
Article 50 addresses defined transparency situations for providers and deployers, subject to its scope and exceptions. The Commission describes duties concerning direct interaction with natural persons, machine-readable marking of certain generated or manipulated outputs, notification around emotion-recognition or biometric-categorisation systems, and disclosure for certain deepfakes and public-interest text.
For management, the relevant outcome is not a generic label placed on every use of AI. It is a defensible view of which public interaction or output engages the rule, which legal entity carries the relevant role, and whether product, communications and third-party arrangements tell the same story.
05 / The revised high-risk dates are category-specific
Regulation (EU) 2026/1744 moved the application dates for the Chapter III high-risk framework. Article 6(2) and Annex III systems now point to 2 December 2027; Article 6(1) and Annex I product-related systems point to 2 August 2028. The categories cover different risk routes, so the later date cannot be applied across an organisation’s entire AI estate as one blanket assumption.
The change gives standards, authorities and affected organisations more time for the high-risk framework. It does not displace other applicable AI Act provisions, product rules, data-protection duties, consumer law, sector regulation or contractual commitments. Those boundaries require current, fact-specific legal analysis.
06 / Role clarity now affects commercial credibility
The AI Act distinguishes providers, deployers, importers, distributors and providers of general-purpose AI models. A company can occupy different roles across products and use cases, and a supplier’s description does not automatically settle the customer’s position. This is why AI governance becomes a leadership issue when procurement, product design, public communications and regulated operations describe the same system differently.
A credible management position connects the active obligation, the entity and role, the relevant deployment, accountable ownership and the point at which specialist advice is required. It should support a decision about continuing, changing or pausing a commercial commitment without turning the regulatory timetable into a false assurance.
07 / Why it matters
The management question is no longer whether every AI deadline falls on the same day. Leadership must distinguish obligations that apply now from high-risk rules that moved, while maintaining one accountable view of organisational roles, deployed systems, public-facing AI interactions and third-party model exposure. The useful outcome is a defensible decision position: what is active, what remains classification-dependent, who owns each exposure and when qualified legal advice is required.
08 / Sources
- European Commission: Navigating the AI Act ↗
- European Commission: Article 50 transparency guidelines ↗
- European Commission: general-purpose AI obligations ↗
- EUR-Lex: Regulation (EU) 2026/1744 ↗
This briefing provides general decision context, not legal advice. Applicability depends on the organisation’s role, the system and the use case; verify the current law and obtain qualified advice for the specific facts.