On August 2, 2026, Article 50 of the EU AI Act became generally applicable and enforceable by national authorities.
It is not a future milestone. It is live, and it applies whether or not your system is classified as high risk.
For crypto platforms, most of which now run AI support agents or generate marketing content with models, the relevant question is not whether the rule applies. It is whether anyone checked.
The Four Situations Covered
Article 50 splits duties between providers, who build and place a system on the market, and deployers, who use one under their own authority.
Systems that interact directly with people must tell users they are dealing with AI. That covers support chatbots and voice agents.
Systems generating synthetic audio, image, video or text must mark outputs in a machine-readable format so they are detectable as AI-generated.
Deepfakes must be labeled. So must certain AI-generated text published on matters of public interest.
Emotion recognition and biometric categorization systems must disclose their use to the people subjected to them.
Narrow carve-outs exist for law enforcement uses authorized by law, minor assistive editing, and certain artistic or editorial contexts.
The Reach Is Extraterritorial
This is the part non-EU businesses get wrong.
The AI Act applies to providers, deployers, importers and distributors that place AI on the EU market, and also where AI outputs are used within the EU.
A platform based in Dubai, Singapore or the United States with European users is in scope. Being headquartered elsewhere is not an exemption.
Non-compliance carries administrative fines of up to 15 million euros or 3% of worldwide annual turnover, whichever is higher. Proportionality is applied for SMEs and small mid-caps.
| Date | What applies |
| August 1, 2024 | AI Act entered into force |
| February 2, 2025 | Prohibitions and AI literacy obligations |
| August 2, 2025 | General-purpose AI model obligations |
| August 2, 2026 | Article 50 transparency obligations, now live |
| December 2, 2026 | Marking deadline for pre-existing generative systems |
| December 2, 2026 | Ban on AI generating non-consensual sexual content and CSAM |
| December 2, 2027 | Standalone high-risk systems, delayed |
| August 2, 2028 | High-risk systems embedded in products, delayed |
What Got Delayed, and What Did Not
August 2 was originally billed as the date the high-risk provisions took effect. That did not happen.
The AI Omnibus, Regulation (EU) 2026/1744, came into force on July 27, 2026 and pushed standalone high-risk obligations to December 2, 2027, and high-risk systems embedded in regulated products to August 2, 2028.
The stated reason is standardization. The EU’s standards bodies need time to publish the technical standards that would give providers a presumption of conformity.
The transparency rules were not delayed. That distinction has been widely misread, with coverage treating the Omnibus as a general reprieve.
The Omnibus also moved faster in one direction. From December 2, 2026 it bans AI systems that generate non-consensual sexually explicit content or child sexual abuse material.
The One Genuine Grace Period
There is a single narrow transition. Generative AI systems already on the market before August 2 have until December 2, 2026 to comply with the machine-readable marking and detection requirement under Article 50(2).
Everything else applies immediately to all in-scope systems regardless of when they launched.
Content generated and published before August 2 does not need retroactive labeling.
The Code of Practice Route
The AI Office has published a voluntary Code of Practice on Transparency of AI-Generated Content, developed by independent experts through a multi-stakeholder process. It includes a set of icons for labeling.
Signing is optional. The underlying obligations are legal requirements either way.
What signatories get is a degree of presumption of conformity and a more favorable enforcement posture. Non-signatories face closer scrutiny and have to demonstrate compliance by other means.
Several major AI providers have signed. For a smaller platform without a compliance function, this is the cheapest available path to a defensible position.
The Commission also adopted detailed guidelines on Article 50 scope and application on July 20, 2026.
Why Crypto Firms Should Read This Twice
This industry has been through an EU compliance cycle recently, and the pattern was not encouraging.
Optimisus covered how the MiCA licensing deadline forced much of the industry out of Europe when firms left preparation too late.
MiCA itself was framed as the standard-setter when the EU approved it, and regulators elsewhere have followed with their own tightening, including Dubai’s VARA overhauling AML rules for every licensed firm.
Article 50 is a lighter obligation than MiCA licensing. It is also live now rather than approaching, and enforcement sits with national market surveillance authorities rather than a single central body.
The Practical Checklist
Identify every system that talks to users, generates content, or infers emotion or biometric category. That inventory is the whole job in most organizations, and it usually turns up more systems than expected.
Add clear disclosure wherever a user is interacting with AI, delivered in an accessible and timely way rather than buried in terms of service.
Talk to your model provider about machine-readable marking. Most platforms will not implement provenance marking themselves and will depend on whoever supplies the model.
Decide on the Code of Practice. If nobody at the company owns AI compliance, signing is a reasonable default.
None of this is expensive. Discovering in an enforcement action that a customer-facing agent never identified itself is.
Sources
- European Commission, Safer and more transparent AI, August 2, 2026 — https://commission.europa.eu/news-and-media/news/safer-and-more-transparent-ai-2026-08-02_en
- Goodwin, Not delayed, not deferred, EU AI Act transparency obligations are now in force — https://www.goodwinlaw.com/en/insights/publications/2026/08/alerts-technology-dpc-eu-ai-act-transparency-obligations-now-in-force
- Cooley, EU AI Act transparency obligations take effect 2 August 2026 — https://www.cooley.com/news/insight/2026/2026-08-03-eu-ai-act-transparency-obligations-take-effect-2-august-2026
- Help Net Security, EU begins enforcing AI Act, putting AI models under the microscope — https://www.helpnetsecurity.com/2026/08/04/eu-ai-act-enforcement-ai-models/
- European Commission, Code of Practice on Transparency of AI-generated Content — https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content
- Data Protection Report, The EU AI Act, when does it become enforceable now — https://www.dataprotectionreport.com/2026/07/the-eu-ai-act-when-does-it-become-enforceable-now/
Optimisus covers crypto and technology news for readers who want the detail behind the headline.
