EU AI Act Enforcement Begins: Chatbots Must Disclose, Deepfakes Must Be Labeled
Summary: Europe's AI Act moved from policy to enforcement on August 2, flipping on transparency rules, high-risk AI obligations, and the EU AI Office's investigative powers in a single day.
Key Facts
- Disclosure mandate: Interactive AI systems — chatbots, voice assistants — must now inform EU users they are speaking with an AI, not a human.
- Watermarking required: Deepfakes and AI-generated or altered content must carry machine-readable markers so they can be detected automatically.
- High-risk obligations live: Requirements for AI deployed in hiring, credit scoring, healthcare, and law enforcement officially took effect on the same date.
- Penalties: Violations risk fines up to €15 million or 3% of worldwide annual turnover, whichever is higher.
Why It Matters
This is the first time a major jurisdiction has switched on binding, enforceable rules for general-purpose AI models at scale. The EU AI Office can now demand documentation, access models, order corrective measures, and levy fines against providers like OpenAI, Anthropic, and Google — all of whom serve European users. The practical compliance burden falls hardest on companies whose products blur the line between AI and human interaction, making this a watershed moment for how AI is deployed globally.
Further Reading
- European Commission press release — European Commission
- Enforcement deep-dive — Help Net Security