2 August 2026 is the next hard deadline in the EU AI Act calendar. While the high-profile obligations for high-risk AI systems do not kick in until December 2027, August 2026 introduces a set of transparency requirements that apply to a far wider range of businesses — including many that do not consider themselves to be in the "AI industry" at all.
Key date: 2 August 2026 — Article 50 transparency obligations enter into force. Non-compliance from this date can result in fines of up to €15 million or 3% of global annual turnover.
What Article 50 Actually Requires
Article 50 of the EU AI Act establishes transparency obligations for four categories of AI system:
1. Chatbots and conversational AI
Any AI system designed to interact with natural persons through text, voice, or other means must clearly disclose that it is an AI — unless the context makes this obvious. This applies to customer service chatbots, AI assistants on websites, and virtual agents in any sector.
The disclosure must be made at the beginning of the interaction, not buried in terms and conditions.
2. AI-generated or AI-manipulated content (deepfakes)
AI systems that generate images, audio, video, or text that could be mistaken for real content must label that content as artificially generated or manipulated. This covers:
- Marketing images generated by AI tools
- AI-generated video or audio content
- Synthetic voices used in customer communications
- AI-written content presented as human-authored (in some contexts)
3. Emotion recognition and biometric categorisation
Operators of AI systems that detect emotions or categorise people by protected characteristics (ethnicity, political opinion, etc.) must inform individuals when such a system is being used on them.
4. Text generated to inform the public
AI systems used to generate text for public information purposes — news articles, government communications, public health messaging — must label the content as AI-generated.
Who Is Affected?
Article 50 applies to any company that deploys these AI systems in the EU, not just companies that build them. That means:
- E-commerce — AI chat support, personalised product descriptions
- Marketing agencies — AI-generated visuals or copy
- HR platforms — AI interview tools, sentiment analysis
- Media companies — AI-written or AI-assisted editorial content
- Customer service operations — any AI-powered contact channel
- Healthcare providers — AI symptom checkers or patient communication tools
If you use any of these tools — even as a SaaS product from a third-party vendor — you are responsible for ensuring the transparency disclosure is in place.
What Is Excluded?
Article 50 includes a narrowly defined exception: AI systems used for purposes authorised by law for the prevention, detection, investigation, and prosecution of criminal offences. Standard business use cases are not exempt.
There is also an exception where the disclosure "would not be necessary" given the obvious context — but this is intended for scenarios where a reasonable person would already understand they are interacting with AI, not as a general opt-out.
What You Need to Do Before August 2026
- Audit your AI touchpoints — map every AI-powered tool that interacts with customers, generates content, or analyses human behaviour.
- Check your vendor contracts — if a SaaS provider is running the AI, confirm in writing who is responsible for the transparency disclosure.
- Add disclosure notices — update chatbot opening messages, add labels to AI-generated content, and review website and app UX to surface disclosures clearly.
- Document your compliance — keep records of what systems you use, how they are disclosed, and when you reviewed them.
- Train your teams — marketing, customer service, and product teams need to understand what "AI-generated" means in the context of their work.
The AI Omnibus: What Changed in May 2026
The provisional agreement reached in May 2026 (the "AI Omnibus") did not change the August 2026 transparency deadline. What it did change was the high-risk AI deadline: obligations for Annex III systems (recruitment, credit, law enforcement, etc.) were deferred from August 2026 to December 2027. The transparency obligations in Article 50 remain on their original schedule.
This means August 2026 is now the near-term compliance priority for the vast majority of businesses.
Key Takeaway
August 2026 is closer than most companies think, and the transparency obligations apply to almost any business using AI to interact with customers or generate content. The good news: compliance is achievable without a major technical overhaul — it is primarily about disclosure, documentation, and vendor accountability. The time to act is now.
Not sure where you stand?
Our free assessment tool maps your AI systems to the relevant EU AI Act obligations in minutes.
Start free assessment