The bottom line: Disclosure requirements for AI systems with user interaction become mandatory from August 2026, but the core compliance risk lies in the lack of control over AI deployment in existing business processes.
On 2 August 2026, the transparency obligations from Article 50 of the EU AI Act take effect: chatbots must disclose themselves as automated systems, AI-generated or manipulated content requires technical identifiers. Violations carry substantial fines.
Article 50 of the EU AI Act becomes binding on 2 August 2026. Automated systems such as chatbots must disclose that they are machines – unless this is obvious. Deepfakes and manipulated image, audio and video content must be equipped with a technical identifier, such as machine-readable markers. Similarly, AI-generated text on matters of public interest must be labelled as such, unless a human has editorially verified it. In Germany, the Bundesnetzagentur monitors compliance; at EU level, the AI Office does.
Not all planned measures are being implemented in full. The obligation for digital watermarks for generative AI providers has been postponed to December 2026. For chatbots and generative systems already in use, there is a transition period until 2 December 2026. The delays for high-risk AI systems are significantly more far-reaching: through the EU Digital Omnibus that took effect on 27 July 2026 (Regulation 2026/1744), the regulation of high-risk systems under Annex III has been postponed by one year to December 2027, and for AI systems embedded in vehicles or medical devices under Annex I even to 2028.
The AI Act has been activated in phases since August 2024: in February 2025, bans on certain AI practices and competence requirements came into force; since August 2025, governance rules and requirements for providers of general-purpose AI (GPAI) models have applied. On 2 August 2026, the most comprehensive step to date follows: transparency obligations for virtually all providers and operators of AI systems with direct user contact.
However, the disclosure requirement addresses only the visible side of compliance. The core question for companies is rather: at which points in business processes is AI already being used, and who is responsible for it? A label does not answer how content was created, whether it underwent human review and who is ultimately liable. Governance structures and clear responsibilities must be established from the outset, not retrofitted as compliance measures afterwards. Regulatory competition will be decided less by speed and efficiency than by who builds trust through traceable controls.
Source: www.it-daily.net · Published 31 July 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.3.