Policy & Regulation

• 

  After 2 August 2026: What AI Images and Videos Must Disclose

  A practical guide to EU AI Act Article 50, visible deepfake labels, machine-readable marks, watermarking, and signed media provenance.

  August 26, 2026

• What Content Safety Actually Looks Like in Code

  Policy documents say what is prohibited. Code has to decide, per string, at build time. Notes from wiring content rules into a generative media pipeline: why the linter must be context-blind, why an LLM is the wrong last line of defence, and the categories that are risky without containing a single prohibited word.

  August 1, 2026
• 

  What "AI Watermarking" Actually Means — and Why It's Harder Than It Sounds

  C2PA metadata and invisible signals like SynthID both promise to mark AI content — how each actually works, and why screenshots and compression defeat them.

  July 15, 2026