EU enforces AI transparency rules with €15M fines

The European Union's artificial intelligence transparency rules took effect on August 2, 2026, compelling publishers to label AI-generated content and inform users when they engage with an AI system.

Under Article 50 of the AI Act, content must carry an "AI" mark if machines assisted its creation, or "AI-Generated" if fully produced by AI, according to European Commission guidelines. Examples include fully AI-generated deepfake videos of politicians, AI-composed music, and AI-generated news summaries. Users must also be told when they are chatting with a bot rather than a human.

"People should know when they are interacting with AI or exposed to AI-generated content," the Commission said, explaining the obligations are intended to foster trust and integrity in the information ecosystem. The rules extend to AI systems that recognize emotions or process biometric data.

Providers of general-purpose AI models face separate requirements to document information, publish training-data summaries and implement copyright policies. Non-compliance can trigger national enforcement with fines up to €15 million or 3% of global annual revenue for companies, and €750,000 for EU bodies, with proportionality considered for small and mid-sized businesses.

Personal communications and "evidently artistic" satirical or fictional works are exempt. The Commission, which will oversee compliance, called effective implementation "a key priority."

The industry has criticized the rules as an undue burden that could slow innovation. But the Commission argued they create a single market with harmonized rules, simplifying development while addressing risks to health, safety, fundamental rights, democracy and the rule of law.