Verification is a field full of confident numbers and quiet caveats. We write the caveats down. Expect at least two posts a month on provenance, watermarks, detector reliability, and the standards that govern them.
The single most important thing to understand about AI-content verification — and the reason Provenote leads with provenance, not a detector score.
In August 2026 Anthropic began watermarking Claude’s text output. Here is a plain-language read on how it works, what it proves, and where its limits sit.
A detector’s advertised accuracy is the wrong number. Here’s the number that actually matters, what it costs a real person when it’s wrong, and how Provenote grades every detector it touches.
The EU AI Act’s transparency duties have applied since 2 August 2026 — not December. Here’s what has to be marked, what has to be visibly disclosed, where a hidden watermark isn’t enough, and what the December 2026 date really is.
C2PA is the “nutrition label” for digital media, and it now reaches into documents. But there’s one thing it fundamentally can’t do for writing — and knowing that line is the whole point.
A missed detection costs you a little certainty. A false positive costs someone else their credibility. Here is the arithmetic almost nobody runs before switching a detector on.
Article 50 makes you say you used AI. That is a different thing from being able to prove how something was made — and the difference decides who ends up protected.
An author drafts with AI, then edits it into their own voice. Can anyone detect that? No. Here is exactly why, in the model builders’ own words, and what follows from admitting it.