EU Mandates New AI Labeling Rules: Chatbots and Deepfakes Must Be Marked by August
Published on 07/29/2026 at 09:22 | Redaktion boerse-global.de
Businesses across Europe face a tight deadline to comply with sweeping transparency requirements for artificial intelligence, as new labeling rules take effect on August 2, 2026. The regulations, drawn from Article 50 of the EU AI Act, force companies to clearly disclose when customers or employees are interacting with AI systems — whether through chatbots, deepfakes, or AI-generated text.
Under the rules, any organization using chatbots for customer service or internal communication must inform users about the automated nature of the conversation. AI-generated or substantially altered content — known as deepfakes — must carry machine-readable markers or watermarks that identify them as artificially produced.
Providers of AI solutions are required to embed tamper-proof metadata into their outputs. Anyone distributing AI-written text on public-interest topics also falls under the labeling obligation. There is an exemption for content that undergoes human editorial review before publication, or that only performs supportive functions such as spell-checking.
As you review your AI compliance obligations, remember that workplace safety documentation also demands careful oversight. A free toolkit with 41 ready-to-use templates and checklists helps you manage risk assessments, fire safety, and manual handling with confidence. Download the free Risk Assessment Toolkit
The timeline for compliance includes a grace period for existing AI systems. While the general transparency duties kick in on August 2, older systems have until December 2, 2026, to be adapted. The Digital Omnibus Regulation (EU 2026/1744), which entered force on July 27, 2026, helped clarify the implementation schedule.
Certain uses are exempt from the labeling requirement entirely. Private, non-commercial use is excluded, as are artistic or satirical works that carry no risk of deception. A complete ban applies from December 2 to applications that generate non-consensual pornographic material or child sexual abuse imagery.
Some major technology platforms have already started marking AI-generated content on their sites, moving ahead of the legal mandate.
New regulations aren't just hitting AI — workplace health and safety rules also demand your attention. Over 37,000 UK businesses already use a free toolkit with risk assessments, checklists, and toolbox talks covering COSHH, fire safety, and more. Get the free Health & Safety Toolkit
Non-compliance carries steep financial penalties. Companies that violate the transparency rules face fines of up to 15 million euros or three percent of their global annual turnover. In the credit sector, where credit scoring remains classified as a high-risk application, penalties can reach 35 million euros or seven percent of turnover.
The Digital Omnibus Regulation also pushed back some obligations for high-risk AI systems. Stricter requirements for AI applications in the lending sector will not apply until December 2, 2027, and for AI systems embedded in products, the deadline extends to August 2, 2028. In Germany, the Federal Financial Supervisory Authority (BaFin) remains the lead regulator for the financial sector, while the EU AI Office gains expanded enforcement powers.
Experts advise companies to conduct a comprehensive AI inventory and boost AI literacy among staff. Organizations using integrated solutions such as SAP’s AI assistant “Joule” must determine whether they act as providers or operators, and assess which risk classifications apply to their applications. The administrative burden is expected to be significant, but the cost of inaction is far higher.
Disclaimer regarding our articles: No investment advice, no buy or sell recommendation. Information on prices, companies, and markets is provided without guarantee; changes are possible at any time. Stock market transactions can lead to substantial losses. Our articles are created and reviewed in whole or in part automatically with the support of AI.
