Europe's transparency rules go live: chatbots must say they're AI, deepfakes must carry labels
As of August 2, interactive AI systems in the EU must tell users they are talking to a machine, and realistic synthetic media must be labeled and watermarked. Alongside the Commission's new enforcement powers, it is the first continent-wide rule making AI disclosure a legal default rather than a design choice.
The disclosure rule is philosophically bigger than it looks. For the first time at continental scale, the law asserts that humans have a right to know when they are talking to software — not as platform policy, not as best practice, but as an enforceable default with the AI Office and national authorities behind it. Every chatbot, voice agent, and synthetic-media pipeline serving EU users now carries that duty.
The timing collides with the agent boom deliberately. The same week these rules activated, consumer agents that phone real stores and enterprise voice agents that answer business lines are going mainstream — conversations where neither side may be human. Europe's answer is that the machine must say so; the open question is how a disclosure regime built for chatbots maps onto agent-to-agent commerce nobody watches.
Enforcement gives the rule teeth the 2025 obligations lacked. With the Commission now empowered to act against general-purpose model providers and penalties scaling to serious percentages of global turnover, transparency compliance moves from the ethics deck to the risk register. The era in which passing as human was a product feature is, in Europe at least, legally over.
European Commission — Commission starts enforcing AI Act rules and new transparency requirements on 2 August → · EU Digital Strategy — Commission starts enforcing AI Act rules and transparency requirements →