August 2, 2026 is the day Article 50 of the EU AI Act stopped being a compliance deadline on a roadmap and became active law, alongside the Commission's enforcement powers over general-purpose AI providers kicking in the same day after a one-year grace period that started August 2025. Unlike the Act's high-risk category, which only touches narrow use cases like hiring or credit scoring, Article 50 is written broadly: it applies to essentially any AI system that talks to, writes for, or presents synthetic media to a person inside the EU. Concretely, if you're building a chatbot, virtual assistant, or any conversational interface, you now have a legal obligation to design it so users are told they're interacting with AI rather than a human, unless it's obvious from context. If your system generates or manipulates audio, image, video, or text content, that output needs to be marked as AI-generated in a machine-readable format, and if you're deploying anything that produces deepfakes or AI-written text about matters of public interest, you need explicit disclosure. None of this requires exotic engineering, but it does require a pattern most teams haven't built yet: a disclosure layer baked into the product rather than bolted on. Practically, that means things like a persistent "you're chatting with AI" indicator in conversational UI, embedding C2PA-style provenance metadata or visible watermarks in generated media pipelines, and logging what content your system produced so you can demonstrate compliance if asked. Penalties for getting this wrong are real: up to €15 million or 3% of global annual turnover, whichever is higher. If you ship anything AI-facing to EU users, this is the week to check whether your product's disclosure UX is more than an afterthought, because the rule is no longer aspirational.