EU AI Act Article 50 Takes Effect, Requiring Chatbots and Synthetic Media to Identify Themselves

Article 50 of the EU AI Act became enforceable on August 2, 2026, requiring AI-powered chatbots and synthetic media to clearly disclose their non-human nature to users across the European Union.

The rule mandates that any AI system interacting with humans in real time must inform users they are communicating with an AI, not a person. It also requires that AI-generated images, audio, and video be labeled as synthetic content.

The regulation applies to companies operating in the EU regardless of where they are headquartered. U.S. tech companies including Google, Meta, and OpenAI must comply or face fines of up to 15 million euros or 3 percent of global annual revenue, whichever is higher.

California's SB 942 AI Transparency Act also became operative in August 2026, requiring large-scale generative AI providers to embed C2PA-compatible provenance data in their media. C2PA is a technical standard that allows users to verify the origin and authenticity of digital content.

The two regulations represent the most significant AI transparency requirements to take effect anywhere in the world. Advocates say they will help combat misinformation and protect consumers from being deceived by AI-generated content.

Tech companies have spent months preparing for compliance. Some have added disclosure banners to chatbot interfaces. Others have integrated watermarking tools into their image and video generation systems.

Critics argue the rules are difficult to enforce and that bad actors will find ways to strip watermarks or avoid disclosure requirements. Regulators say enforcement will focus on large platforms and repeat violators.

The EU AI Act was passed in 2024 and has been rolling out in phases. Additional provisions covering high-risk AI systems in healthcare, education, and law enforcement are scheduled to take effect in 2027.