What August 2 actually requires
Article 50 of the EU AI Act establishes four categories of transparency obligation. Not all four apply equally to every business. The most relevant for small and medium businesses are these three. First, chatbot and AI interaction disclosure. Any AI system designed to interact directly with people must clearly inform users that they are communicating with an AI, not a human. This disclosure must happen at the start of the first interaction, in a clear and distinguishable way, not buried in terms of service or fine print. If your business uses an AI chatbot for customer service, sales support, website chat, appointment booking, or any other customer-facing conversation, EU users must be informed they are talking to AI before the conversation proceeds. Second, AI-generated content labeling. If your business deploys AI to create or manipulate images, audio, video, or text and publishes that content publicly, disclosure obligations apply. For deployers, the requirement covers AI-generated deepfakes and AI-generated text published to inform the public on matters of public interest. These rules apply from August 2. Third, the machine-readable marking requirement. Providers of generative AI systems must technically mark outputs as AI-generated. For systems already on the market before August 2, there is a transitional window to December 2, 2026. For new generative AI systems launched after August 2, the marking obligation applies immediately. What is not required on August 2: the stricter obligations for high-risk AI systems, including documentation, conformity assessments, and quality management systems, have been postponed under the EU's Digital Omnibus proposal. Those take effect December 2027 at the earliest. If your business uses AI in employment decisions, credit evaluation, or other high-stakes applications, that work is important, but August 2 is not the relevant deadline for that category.
Who this actually applies to
The EU AI Act has extraterritorial reach. It applies to any organization that offers AI-enabled products or services in the EU market, regardless of where that organization is based. US companies are not exempt. The practical question is: can EU residents access your AI-powered tools or AI-generated content? If you operate an e-commerce site, a SaaS product, a professional services platform, or any other digital presence with AI-powered features, and EU residents can reach it, you are likely in scope for the transparency obligations. The obligations apply to two types of actors. Providers are the companies that develop and supply AI systems. Deployers are the companies that put those systems to use in customer-facing applications. For most small and medium businesses, the relevant role is deployer. That means even if you are using a third-party chatbot tool built by a vendor, you as the deployer may carry responsibility for ensuring EU users receive the required disclosure. The Act does not have an SMB exemption for Article 50 obligations. Scope is determined by whether you operate AI-powered tools that EU residents interact with, not by your company size.
What business owners often misunderstand about the delay
The most common misunderstanding is conflating the high-risk AI delay with a general EU AI Act delay. Media coverage has focused on the postponement of high-risk AI system requirements under the Digital Omnibus agreement, reached on May 7, 2026 between the European Parliament and the Council. That coverage is accurate. The detailed obligations for high-risk AI systems in employment, biometrics, credit, and infrastructure were originally due August 2, 2026, and they have been pushed back. But Article 50 transparency obligations are not high-risk system requirements. They sit in a separate part of the Act, and the delay provisions in the Digital Omnibus do not extend to them. Multiple law firm compliance alerts published in June 2026, including from Sidley Austin and Latham and Watkins, confirm that chatbot disclosure obligations apply from August 2 without delay. The second misunderstanding is about what triggers the obligation. The disclosure requirement does not depend on whether your AI tool is sophisticated. A basic automated response system that holds text conversations with users may qualify as an AI system designed to interact with natural persons under the Act. If EU users cannot immediately tell from context that they are talking to software, the disclosure is required. The third misunderstanding concerns penalties. Violations of Article 50 can result in fines of up to 15 million euros or 3 percent of total worldwide annual turnover, whichever is higher. These are not numbers reserved for large enterprises. A company with 2 million euros in annual revenue faces a potential 60,000 euro penalty for a transparency violation. Whether enforcement actions target smaller companies first is a separate question, but the legal exposure is real from day one.
What a serious business should do before August 2
Five weeks is not long. The starting point is an honest assessment of scope. The first question is whether your business serves EU customers or whether EU residents can access your digital products or services. If the answer is genuinely no, and you have no EU-facing presence, the August 2 obligations do not apply. That is a legitimate answer, but confirm it rather than assume it. If the answer is yes or uncertain, the next questions are straightforward. Do any customer-facing interactions use AI, including chatbots, automated assistants, AI-powered search, or any system that responds to users in conversation format? Does your business publish AI-generated content that EU users can access, including marketing materials, product descriptions, or informational content produced using generative AI tools? For each yes: does the current system or content already include a clear, upfront disclosure that AI is involved? Adding chatbot disclosure is usually not technically complex. It is often a matter of updating the chatbot configuration to show a disclosure message at the start of conversations, or reviewing the settings of a third-party tool to confirm disclosure is active. The harder task is confirming who holds responsibility when AI tools are provided by a third party. If you use a vendor's chatbot platform on your website, the vendor agreement may or may not address EU AI Act compliance. Review the agreement to confirm whether the vendor's system already meets the disclosure requirements and which party bears responsibility for ensuring users are informed. If the contract is silent on this, that is a gap worth closing before August 2, not after. For businesses with any uncertainty about their scope or obligations, the right step is to work with a qualified legal advisor who can assess the specific situation. This post provides business awareness, not legal advice.
The Atlacis view
The EU AI Act is often framed as a large-company concern. The practical experience of August 2 is going to be different. Many small and medium businesses use AI chatbots, AI customer service tools, or AI-generated content without having thought carefully about what that means for their EU-facing operations. The transparency obligations under Article 50 are not the most complex part of the EU AI Act. But they are the first part that requires real action by real businesses, and the window is five weeks. The pattern we see with business owners and AI compliance is the same as with data privacy: the owners who handle it well are the ones who understood their exposure before the deadline, not the ones who scrambled after. The practical work here is manageable for most businesses. Map which AI tools interact with EU customers. Confirm that those tools include the required disclosures at the start of interactions. Review vendor agreements to understand who is responsible for compliance. For most businesses with a modest EU presence, this is a week of focused effort, not a months-long compliance project. What makes it harder is starting at week four instead of week one. The time to understand your AI exposure is before the deadline, not after. At Atlacis, we help business owners understand where their AI exposure is, including regulatory exposure, before it becomes a compliance problem. If you are not certain whether August 2 applies to your business, or which AI tools are in scope, that question is worth resolving now.
The short version
- Article 50 of the EU AI Act imposes transparency obligations that take effect on August 2, 2026. The most immediate requirement is that AI chatbots and AI interaction systems must clearly disclose to EU users that they are interacting with AI, at the start of the first conversation.
- The high-risk AI system obligations have been delayed to December 2027 or later under the EU Digital Omnibus proposal. The chatbot disclosure rules in Article 50 were not delayed and apply August 2.
- The EU AI Act applies to any organization offering AI-enabled products or services in the EU market, including US companies with EU-facing websites, apps, or customer interactions. Location of incorporation is not the determining factor.
- Businesses using third-party AI tools are responsible as deployers for ensuring EU users receive the required disclosures. Vendor agreements should specify who holds the compliance obligation.
- Fines for Article 50 violations can reach 15 million euros or 3 percent of total worldwide annual turnover, whichever is higher. This exposure applies to businesses of any size operating in scope.
- The practical starting point is three questions: do EU residents interact with AI-powered systems in your business, does your business publish AI-generated content that EU users access, and do your current systems already include the required disclosures?
Where ATLACIS can help
Sources
- Sidley Austin: EU AI Act Transparency Obligations: Preparing for Compliance by 2 August 2026 (June 24, 2026)
- Greenberg Traurig: Deepfakes, Chatbots, AI-Generated Text: EU Commission Details Transparency Obligations Under the AI Act (June 2026)
- Latham and Watkins: AI Act Update: EU Resolves to Change Rules and Extend Deadlines
- EU Commission: Code of Practice on Transparency of AI-Generated Content (June 10, 2026)
- EU AI Act Article 50 official text (artificialintelligenceact.eu)