EU AI Act

What it is

The EU AI Act is the world's first broad, binding AI law. Most of its weight falls on "high-risk" systems (see the Annex III page), but the part that touches nearly every AI-flavored SaaS product is Article 50: transparency. As of August 2, 2026, these obligations are enforceable. The grace period is over.

Does it apply to you?

If EU users can reach your product, yes. The Act applies based on where your users are, not where your company is registered; a solo founder in Denver serving customers in Berlin is in scope. The obligations trigger when users interact with an AI system (chatbots, assistants, AI-driven features) without it being obvious, or when your product generates or manipulates text, images, audio, or video presented to people.

What you must do

Two things, and the first costs one sentence of UI copy.

1. Tell users when they're interacting with AI, unless it's obvious from context. A line near your chat widget ("You're chatting with an AI assistant") satisfies the core duty.
2. Disclose AI-generated content. If your product outputs AI-generated media that users share or publish, the output needs to be identifiable as AI-generated. Machine-readable marking becomes mandatory on its own schedule; see the AI Content Marking page.

Penalties

Transparency violations carry fines up to €15M or 3% of global annual turnover, whichever is higher. You'll often see €35M or 7% quoted. That figure belongs to the prohibited-practices tier (social scoring, banned manipulation techniques), not transparency. Getting the tier right matters when you're deciding how urgently to act.

What SaaSalyst checks

The eu_ai_act_transparency check looks for AI-disclosure signals on sites showing evidence of AI features. Since August 2, 2026 it reports a fail rather than a warning for in-scope products with no detectable disclosure.