The EU's AI Act grows teeth: chatbots must now disclose they're machines

The European Union's landmark AI Act entered a new enforcement phase on 2 August 2026, when the @EU_Commission began actively applying transparency rules that require any AI system interacting with people to identify itself as a machine from the very first exchange.

What the Rules Require

Under the new rules, chatbots and other interactive AI systems must tell users they are dealing with AI rather than a human, and deepfakes, including images, videos, or audio edited or generated using AI, must be labelled. AI-generated or altered content must also carry machine-readable marks to facilitate detection.

The rules are intended to reduce fraud, impersonation, misinformation, and other forms of deception as synthetic content becomes harder to distinguish from authentic material. Notably, deepfake rules apply even without intent to deceive: content that looks or sounds like a real person must be labeled even if no deception was intended.

The duties reach businesses wherever they are established, so a provider outside the EU that serves EU users is still caught. That scope extends to crypto trading bots and automated customer support agents used by platforms with European users.

Non-compliance can result in administrative fines of up to 35 million euros or, for larger companies, up to 7% of total worldwide annual turnover, whichever is higher. The framework for fines exceeds even the penalties provided under the General Data Protection Regulation, which are capped at 20 million euros.

High-Risk AI Rules Pushed Back

Not all of the AI Act's provisions kicked in at once. Under the Digital Omnibus, a provisional agreement reached on 7 May 2026 and pending formal adoption, the high-risk AI deadline for Annex III systems, covering areas such as hiring, credit scoring, and law enforcement, was deferred from 2 August 2026 to 2 December 2027.

The Digital Omnibus moved those high-risk obligations to December 2027, and some read that as the AI Act being delayed overall. It is not. Only one track moved. The chatbot disclosure requirements, deepfake labelling, and the Commission's power to investigate and issue fines are all live as of this month.

More than 180 organisations have signed a voluntary Code of Practice designed to help companies apply the transparency rules for AI-generated content. Adherence to the Code is expected to serve as a practical route to demonstrating compliance for many businesses.

Sources:
European Commission: Commission starts enforcing AI Act rules and new transparency requirements
EU AI Act Article 99: Penalties
EU AI Act 2026: Penalties, Risk Tiers and New Deadlines