EU starts enforcing AI Act transparency rules: chatbots and deepfakes must now identify themselves

On 2 August 2026, the European Commission's AI Office and national competent authorities began enforcing the transparency obligations under Article 50 of the EU AI Act. From that date, providers and deployers of covered AI systems must disclose AI involvement in four areas: direct interaction with individuals (for example, chatbots must identify themselves as AI rather than a human), AI-generated or altered content generally, emotion recognition and biometric categorisation systems, and deepfakes or AI-generated text on matters of public interest, which must carry machine-readable marks. A limited transitional period, until 2 December 2026, applies specifically to the marking-and-detection obligation for generative systems already on the market before this date. Non-compliance can be reported by any individual or business, including a competitor, to a national market surveillance authority, which can compel internal documents and impose fines of up to 35 million euros or 7% of global annual turnover for the most serious violations.

Analysis: this is the first EU AI Act enforcement date that touches ordinary AI-tool usage directly, as opposed to the high-risk-system provisions researchers and institutions have been preparing for on a longer timeline. For anyone using AI writing or research tools with an EU nexus, the practical question this raises is not whether an AI system is high-risk, but whether anything published or deployed fails to disclose that a person is talking to, or reading output from, a machine.

What to watch: whether the AI Office issues its first public enforcement action or fine under these specific transparency provisions, which would be the first real test of how the complaint-driven enforcement mechanism functions in practice.

Source: European Commission, https://digital-strategy.ec.europa.eu/en/news/commission-starts-enforcing-ai-act-rules-and-new-transparency-requirements-2-august