EU Transparency Rules Force Companies to Label AI-Generated Content From August
Published on 08/08/2026 at 10:21 | Redaktion boerse-global.de
Businesses across the European Union are facing new compliance duties as of 2 August 2026, when Article 50 of the AI Act took effect. The regulation obliges both providers and operators to disclose when users are interacting with chatbots or encountering machine-generated material. The stated purpose is straightforward: preventing deception in digital spaces.
What the New Obligations Cover
Any AI system that interacts directly with people must now identify itself as artificial. The moment a user starts a conversation, it must be clear whether they are dealing with a human or software. Austrian authorities have already updated their own chatbot models to meet the requirement.
The rules extend well beyond text-based assistants. Images, audio files and video content produced by generative AI need machine-readable markings. Deepfakes face the toughest scrutiny — manipulated footage that falsely presents people or events as genuine. Even AI-written articles touching on matters of public interest cannot escape the labelling requirement.
There are carve-outs, however. Content that has passed through human editorial review is exempt, as are artistic, satirical and fictional works. Law enforcement agencies operate under separate provisions.
Workplace Implications for Employers
The compliance burden is growing for companies that deploy AI in human resources. Whether screening job applicants or monitoring employee performance, organisations must now pay closer attention to data minimisation and maintain meaningful human oversight. Legal experts are advising firms to revise works council agreements, making explicit provisions for data processing, access rights and the boundaries of automated decision-making.
As you review your compliance duties around new AI systems, don't overlook your existing workplace safety obligations. A free Health & Safety Toolkit provides ready-to-use risk assessments and checklists that help UK businesses stay compliant with current regulations and protect employees from harm. Download the free Health & Safety Toolkit
High-risk AI used in personnel management carries particularly demanding obligations. The GDPR remains the baseline — automated individual decisions still require human supervision.
Deadlines and Penalties
Systems already operating before 2 August get a grace period. Existing AI applications must be retrofitted by 2 December 2026. Other parts of the regulation roll out in phases: bans on certain practices, including so-called "nudifiers," take effect in December 2026, while obligations for high-risk systems follow in December 2027.
The financial stakes are considerable. Breaching transparency rules can result in fines of up to €15 million or 3 percent of global annual turnover. For prohibited practices, the ceiling rises to €35 million or 7 percent of turnover.
With new regulations arriving thick and fast, staying on top of compliance paperwork is more important than ever. Over 37,000 UK companies use a free Risk Assessment Toolkit with 41 templates and checklists covering fire safety, manual handling and more — everything you need to document workplace hazards properly. Get the free Risk Assessment Toolkit
Industry Pushback and Technical Doubts
Business associations are complaining about the timing of the EU's guidance, much of which only surfaced in June or July 2026. Preparation time, they argue, was simply insufficient. Several member states also lack the institutional framework — Austria, for instance, has yet to designate a national supervisory authority.
Google, Meta and OpenAI have already embedded machine-readable watermarks into their systems. Yet researchers caution that techniques such as invisible pixel noise can often be stripped away with relative ease. More robust standards, including the C2PA model, need to gain traction faster, they say.
