What Happened on August 2?
On August 2, 2026, key transparency obligations under the EU's landmark Artificial Intelligence Act became mandatory. These rules, specifically outlined in Article 50 of the act, apply to any company that develops or deploys AI systems for users in the European
Union. This isn't just about high-risk AI; the regulations cover a broad spectrum of common applications. This includes requiring chatbots to disclose that they are AI, and compelling providers of generative AI tools to ensure their synthetic video, image, or audio outputs are marked in a machine-readable format so they can be identified as artificial.
The Core Problem: AI's Black Box
At the heart of these regulations is a fundamental challenge with modern AI: the 'black box' problem. Large language models can generate incredibly convincing text, images, and audio, but their internal reasoning is often opaque. This leads to what are known as 'hallucinations'—instances where an AI confidently presents fabricated information as fact. Without a clear way to trace the information back to its origin, users have no reliable method to distinguish between truth and sophisticated fiction. This dynamic poses a significant threat, fueling the spread of misinformation and eroding public trust. The EU's move is a direct attempt to force accountability into this process.
Defining Source Verification
The term 'source verification' in the context of the AI Act isn't about fact-checking every sentence an AI produces. Instead, it focuses on two main areas: provenance and disclosure. Firstly, providers of general-purpose AI models must now maintain and publish detailed summaries of the data used to train their systems. This includes respecting copyright and data-mining opt-outs. Secondly, for content like deepfakes or AI-generated articles on matters of public interest, there are strict new labelling rules. Deployers must clearly disclose that the content is artificial unless it has undergone human review where a person takes editorial responsibility. This creates a chain of accountability, from the training data to the final output.
Who Is on the Hook?
The rules have a broad, extraterritorial reach. They apply not only to companies based in the EU but to any organisation whose AI systems or their outputs are used by people within the Union. This means a tech company in India, the US, or anywhere else must comply if its chatbot interacts with EU customers or its AI-generated marketing images are displayed there. The obligations are split between 'providers' (those who build the AI) and 'deployers' (those who use it). Penalties for non-compliance are severe, with fines reaching up to €15 million or 3% of a company's global annual turnover for violating these transparency rules, and even higher for other breaches of the Act.
The Global Ripple Effect
The EU AI Act is widely seen as setting a global benchmark for tech regulation, a phenomenon often called the 'Brussels Effect'. Just as the GDPR data privacy law became a worldwide standard, these AI rules are expected to influence policy far beyond Europe's borders. For Indian IT and SaaS companies with clients in the EU, compliance is now a non-negotiable cost of doing business. More broadly, the Act is transforming skills in demand, creating a need for professionals who blend technical AI knowledge with regulatory expertise. By forcing a conversation around AI transparency and data provenance, the EU is shaping the future of responsible AI development globally.
