Field notes from the proof layer. What we publish for the teams operating under AI Act, GDPR, and state-level synthetic-content law.
Article 50 obligations explained, the five marking duties, and the cost of non-compliance for teams that ship AI-generated content into the European market.
Probabilistic detection produces false positives at scale. Cryptographic watermarks do not. The case for mathematical proof over statistical resemblance.
If you use AI to generate any part of your content, you now have a legal obligation to disclose it. Here is what changed on August 2.
How a European fashion house moved from quarterly IP audits to per-asset trademark-backed certificates, using Shield's brand tier.
Seven months in production. Twenty million certificates. Three lessons we did not see coming. The shape of what comes next.
Six cryptographic operations between the moment you hit generate and the moment your certificate is publicly verifiable. Here is what each one does.