A New Era of Disclosure
The European Union has activated a critical part of its sweeping AI Act, specifically the transparency obligations outlined in Article 50. As of August 2, 2026, companies operating in the EU, or whose services are used by people in the EU, must now clearly
inform users when they are interacting with an AI system. This means that chatbots, virtual assistants, and other AI agents can no longer masquerade as humans. The disclosure must be clear and provided at the start of the interaction, not buried in lengthy terms of service agreements. This move is designed to combat deception, fight misinformation, and empower consumers to make informed decisions about their digital interactions.
What the Rules Mandate
The new rules are multifaceted and apply to both the 'providers' who build AI and the 'deployers' who use them. For chatbots and other interactive AI, the core rule is simple: tell people they are talking to a machine unless it's already obvious. Beyond chatbots, the regulation targets AI-generated content, often called 'deepfakes'. Any synthetic or manipulated image, audio, or video that appears authentic must be labelled. The same applies to AI-generated text on matters of public interest, unless a human editor has reviewed it and taken responsibility. To aid this, providers of generative AI must ensure their output is marked in a machine-readable format, making it easier to detect as AI-generated.
The Compliance Challenge for Tech
For global technology companies, this presents a significant operational test. The rules have a broad reach, affecting any company whose AI outputs are used within the EU, regardless of where the company is based. Failure to comply carries substantial penalties, with fines reaching up to €15 million or 3% of a company's global annual turnover, whichever is higher. To help businesses comply, the EU's AI Office has published guidelines and a voluntary Code of Practice, which over 180 organisations had already signed by the end of July 2026. While new systems must comply immediately, a short grace period until December 2, 2026, is granted for the technical marking of generative systems that were already on the market before the rules came into effect.
Why This Matters Beyond Europe
While these are EU regulations, their impact will be felt globally, creating what is often called the 'Brussels effect'—where companies adopt EU standards worldwide for simplicity. Users in India and elsewhere who use products from Meta, Google, Microsoft, and others will likely start seeing these new disclosures and labels. The AI Act is not just about labelling; it's about setting a precedent for AI governance. The rules aim to foster trust and integrity in the digital ecosystem at a time when distinguishing between human and machine-generated content is becoming increasingly difficult. Other countries, like Canada, are already looking to the EU's framework as a model for their own AI regulations.
