EU AI Act transparency rules take effect requiring chatbot and deepfake labeling

New transparency rules under the EU's AI Act are now in effect, requiring companies to disclose when users interact with AI and to label AI-generated content like deepfakes. The rules, which took effect August 2, 2026, aim to combat manipulation and fraud. Non-compliance can lead to fines of up to €15 million. Existing AI systems have until December 2, 2026, to comply with the new obligations.

Article 50 transparency obligations

• ▪The rules require companies to inform people when they are interacting with an AI system and to label AI-generated content.
• ▪The European Commission's concerns driving these rules include manipulation at scale, fraud, impersonation, and consumer deception.
• ▪Article 50 of the EU AI Act, which sets transparency obligations for AI providers and deployers, came into effect on August 2, 2026.
• ▪New AI systems must comply immediately, while systems launched before August 2, 2026, have a grace period until December 2, 2026.

Provider technical requirements

• ▪AI providers must design systems to explicitly notify users when they are interacting with an AI, unless it is obvious from the context.
• ▪Providers must add a machine-readable mark to synthetic audio, image, video, or text so it can be detected as artificially generated or manipulated.
• ▪Law enforcement systems used for criminal offenses are exempt from the AI interaction disclosure rule, provided third-party rights are protected.

Deployer disclosure duties

• ▪AI-generated text published on matters of public interest must be disclosed, unless a human has reviewed it and holds editorial responsibility.
• ▪Deployers are required to label AI-generated or manipulated deepfake content, including images, audio, and video.
• ▪Deployers of emotion recognition or biometric categorization systems must inform individuals who are exposed to them.
• ▪For artistic, satirical, or fictional works, deepfake disclosures only need to flag the content's existence in a non-disruptive manner.

Enforcement framework

• ▪Failure to comply with the new transparency rules can result in fines of up to €15 million or 3% of a company's global annual turnover.
• ▪Enforcement is split between national market surveillance authorities, the AI Office, and the European Data Protection Supervisor.

Commission implementation guidance

• ▪The European Commission has developed a set of optional, standardized AI disclosure labels that tech platforms can adopt.
• ▪Adhering to the Code of Practice on Transparency of AI-generated Content is one way for companies to demonstrate compliance with marking obligations.
• ▪Companies that do not adopt the Code of Practice must demonstrate compliance through other means deemed adequate by regulators.

2 sources