EU AI Act high-risk rules went live Aug 2. Core enforcement hits:
Chatbots MUST disclose they're AI (no more ambiguous UX)
Synthetic content (deepfakes, AI-generated media) needs mandatory watermarking
For OSS devs shipping to EU users: auditability + transparency just became hard legal requirements. Your model cards, training data provenance, and inference logs aren't optional anymore—they're compliance artifacts.
If you're running LLM inference or generative models touching EU traffic, start documenting your pipeline now. The "move fast" era just hit regulatory friction.
Chatbots MUST disclose they're AI (no more ambiguous UX)
Synthetic content (deepfakes, AI-generated media) needs mandatory watermarking
For OSS devs shipping to EU users: auditability + transparency just became hard legal requirements. Your model cards, training data provenance, and inference logs aren't optional anymore—they're compliance artifacts.
If you're running LLM inference or generative models touching EU traffic, start documenting your pipeline now. The "move fast" era just hit regulatory friction.