The knowledge base
Guides
Engineering-grade, primary-source guidance across the full content authenticity stack. Written by the team that builds these systems, maintained as the standards evolve.
EU AI Act
- EU AI Act Article 50 Explained: Rules, Dates, Penalties EU AI Act Article 50 requires watermarking and disclosure of AI content from 2 August 2026. Full obligations, timeline, and penalties, sourced and dated.
- EU AI Act Watermarking: What Article 50(2) Requires What the EU AI Act's machine-readable marking duty means in engineering terms: techniques per modality, robustness bar, and an implementation checklist.
- AI Content Disclosure Requirements Under the EU AI Act When AI-generated content must carry a visible label under EU AI Act Article 50: deepfakes, chatbots, public-interest text, with concrete UI examples.
- Who Must Comply With the EU AI Act? Providers, Deployers Provider or deployer, EU or not: a decision tree for who owes EU AI Act Article 50 duties, and why US and Canadian SaaS with EU users are in scope.
Watermarking
- AI Content Watermarking: How It Works and How to Build It How AI content watermarking embeds detectable signals in text, images, audio, and video at generation time, and how to build or buy a pipeline.
- SynthID Explained: How Google's AI Watermark Works What Google DeepMind's SynthID watermark actually does across text, image, audio, and video, what it cannot do, and where it fits in a compliance stack.
- Invisible vs Visible Watermarks for AI Content Invisible and visible AI watermarks compared: robustness, user experience, spoofing risk, and which one regulators actually expect under the EU AI Act.
- Can AI Watermarks Be Removed? Attacks and Robustness A taxonomy of AI watermark removal attacks (crop, compress, paraphrase, regenerate, adversarial) and a realistic assessment of what survives in practice.
Content Credentials
- C2PA and Content Credentials, Explained What C2PA and Content Credentials are, how manifests, signing, and verification work, and who backs the standard, explained for product teams.
- Implementing Content Credentials: A Build Guide Step-by-step guide to adding C2PA Content Credentials to a product: SDK choices, signing certificates, manifest design, and gotchas from real builds.
- How to Verify Content Provenance, Step by Step How to check Content Credentials on any image, video, or document, plus what to do when platforms strip metadata or files get re-encoded.
- C2PA vs Watermarking: Which Do You Need? An honest comparison of C2PA provenance metadata and invisible watermarking: strengths, failure modes, and why most real deployments need both.
Deepfake provenance
- Deepfake Provenance: Prove What's Real Why cryptographically signed provenance beats deepfake detection as a long-term trust strategy, how C2PA signing works, and where its limits are.
- Deepfake Detection vs Provenance: Which Holds Up? Side-by-side comparison of deepfake detection and content provenance: real accuracy data, failure modes, coverage, cost, and a verdict per use case.
- How to Verify the Origin of a Photo or Video A step-by-step workflow for newsrooms and trust teams to verify media origin: Content Credentials, metadata, watermark checks, and reverse image search.
Document provenance
- Document Provenance: Proving Authenticity and Integrity What document provenance means, how hashing, signatures, timestamps and audit trails prove authenticity, and when simpler tools are enough.
- Tamper-Evident Documents: Patterns and Tradeoffs How to make documents tamper-evident with hashing, hash chains, append-only logs, digital signatures and RFC 3161 timestamps, with tradeoffs.
- Blockchain Notarization: When Anchoring Hashes Makes Sense How blockchain notarization anchors document hashes for independently verifiable timestamps, and when an RFC 3161 TSA or database is enough.